Asymmetric interactions between the acidic P1 and P2 proteins in the Saccharomyces cerevisiae ribosomal stalk.

The Saccharomyces cerevisiae ribosomal stalk is made of five components, the 32-kDa P0 and four 12-kDa acidic proteins, P1alpha, P1beta, P2alpha, and P2beta. The P0 carboxyl-terminal domain is involved in the interaction with the acidic proteins and resembles their structure. Protein chimeras were constructed in which the last 112 amino acids of P0 were replaced by the sequence of each acidic protein, yielding four fusion proteins, P0-1alpha, P0-1beta, P0-2alpha, and P0-2beta. The chimeras were expressed in P0 conditional null mutant strains in which wild-type P0 is not present. In S. cerevisiae D4567, which is totally deprived of acidic proteins, the four fusion proteins can replace the wild-type P0 with little effect on cell growth. In other genetic backgrounds, the chimeras either reduce or increase cell growth because of their effect on the ribosomal stalk composition. An analysis of the stalk proteins showed that each P0 chimera is able to strongly interact with only one acidic protein. The following associations were found: P0-1alpha.P2beta, P0-1beta.P2alpha, P0-2alpha.P1beta, and P0-2beta.P1alpha. These results indicate that the four acidic proteins do not form dimers in the yeast ribosomal stalk but interact with each other forming two specific associations, P1alpha.P2beta and P1beta.P2alpha, which have different structural and functional roles.

The ribosomal stalk is an important structural element of the large ribosomal subunit directly associated with the interaction of the elongation factors during the protein synthesis elongation step in bacteria (for a review, see Ref. 1). A direct confirmation of this association has recently been shown by cryoelectron microscopy (2,3). Involvement of the stalk components has also been reported in initiation (4,5) and termination (6,7).
Although the data are more scarce, in eukaryotes there is also a relationship between the stalk and the supernatant factors (8 -11), which in the case of Saccharomyces cerevisiae EF-2 has been confirmed by electron microscopy (12). In addition to this function, the eukaryotic stalk, at least the yeast stalk, might participate in a translation regulatory mechanism not reported in bacteria (13).
The bacterial stalk is made of protein L10 and two dimers of proteins L7/L12, the amino-terminal acetylated and nonacety-lated forms of a unique polypeptide. The pentamer L10-((L7/ L12) 2 ) 2 is extraordinarily stable (14) and binds directly to the highly conserved GTPase-related site in the 23 S rRNA (15,16) through the L10 amino-terminal domain (17). Although the three-dimensional structure of the bacterial acidic proteins has recently been resolved (18), the detailed structure of the stalk protein complex is unknown. In fact, the ribosomal stalk is not present in the recently reported 2.3-Å resolution atomic structure of the large ribosomal subunit probably because of its high flexibility (19). Nevertheless a symmetric structure for the pentameric complex has been proposed (20), although the function of the two L7/L12 dimers in the complex might not be the same (21), and a physical separation of both dimers has been proposed (22).
The new role proposed for the ribosomal stalk in eukaryotic organisms correlates with a higher structural complexity and dynamism. Proteins P0 and P1/P2 are the eukaryotic counterparts of L10 and L7/L12, respectively. As in bacteria, one copy of P0 and four copies of P1/P2 seem to form a complex (23). However, the eukaryotic pentamer is less stable and, in contrast to the bacterial pentamer, is readily disassembled by treating the ribosome with ammonium/ethanol buffers (8). The amino acid sequence similarity of the prokaryotic and eukaryotic stalk components is rather low (24). Their structural differences are especially remarkable in the case of P0, which is larger than L10 and contains a carboxyl-end extension of around 100 amino acids not present in the bacterial polypeptide (24). This extension resembles the structure of the acidic P1/P2 proteins, containing a flexible hinge and the same highly conserved 13-amino acid terminal peptide (see Fig. 1). In fact, the P0 carboxyl extension plays the same role as the 12-kDa proteins when these acidic proteins are not present in the ribosome (25,26). As a consequence of this functional similarity, the proteins P1/P2, in contrast to bacterial L7/L12, are not essential for ribosome activity and cell viability (27).
The acidic proteins have also evolved notably. In eukaryotes they are found as a set of genetically independent polypeptides, which can be grouped in two families, P1 and P2, and are made up of a different number of components depending on the species. Only one protein of each type has been found in mammals, insects, and fungi (28 -30), whereas in protozoa (31) and yeast (32,33) there are several. In plants, a third type of acidic protein, P3, has been reported (34).
The presence of several acidic proteins of the same type in the stalk raises a number of questions regarding their respective structural and functional roles. In S. cerevisiae there are four acidic proteins, two of the P1 type, P1␣ and P1␤, and two of the P2 type, P2␣ and P2␤. The estimation of a total of four copies of acidic proteins per yeast ribosome (35) seems to exclude the presence of one dimer of each polypeptide as reported in eukaryotes having only one acidic protein of each type, P1 and P2 (23). There is some evidence suggesting that the acidic proteins are present as monomers in the S. cerevisiae ribosomal stalk (36), although this issue is still not totally settled because at least the P2 proteins are able to form dimers in solution (37). In addition, a number of questions regarding the mutual relationship between different stalk components are still open.
In an attempt to explore further the structural and functional role of the P0 carboxyl-end domain as well as its interaction with the 12-kDa acidic proteins, a series of chimeras were prepared that carry a whole acidic protein replacing the last 112 amino acids forming the original P0 carboxyl terminus. These artificial constructs have been shown to be functional in yeast, and the results obtained from strains expressing them have provided relevant information on the structure of the S. cerevisiae ribosomal stalk.
Yeasts were grown in either YEP medium (1% yeast extract, 2% peptone) or minimal YNB medium (0.67% yeast nitrogen base, 2% carbon source) supplemented with the necessary nutritional requirements. In both cases, the carbon source was either 2% glucose or 2% galactose as required. Escherichia coli DH5␣ was used for the maintenance and preparation of plasmids and was grown in LB medium.

Enzymes and Reagents
Restriction endonucleases were purchased from Roche Molecular Biochemicals, MBI Fermentas, New England Biolabs, and Amersham Pharmacia Biotech and were used as recommended by the suppliers. T4 DNA ligase, calf intestinal alkaline phosphatase, and the DNA polymerase I Klenow fragment were from Roche Molecular Biochemicals. DNA manipulations were performed basically as described previously (39). Polymerase chain reaction was carried out using Pfu DNA polymerase from Stratagene and custom-made oligonucleotides from Isogen following the recommendations of Dieffenbach and Dveksler (40).

Cell Transformations
Bacterial transformations were performed according to the procedure of Hanahan (41). Yeasts were transformed using the lithium acetate method as described previously (42). Plasmids pFL37P0/1␣, pFL37P0/2␣, and pFL37P0/2␤-Using appropriate oligonucleotides as primers, the genes RPP1A, RPP2A, and RPP2B encoding the acidic protein P1␣, P2␣ and P2␤, respectively, were obtained by polymerase chain reaction using Pfu polymerase from plasmids in which they were previously cloned (32,43). At the same time, a new NheI restriction site was introduced downstream from the termination codon (see Fig. 1). On the other hand, taking advantage of the presence in the RPP0 gene of an EcoRV site in the position corresponding to amino acids 203 and 204 and a NheI site in the 3Ј region, the coding sequence encoding the last 112 residues was removed from RRP0 in plasmid BSP0 (38). Afterward the corresponding polymerase chain reaction fragments encoding the acidic proteins were subcloned in the same sites of BSP0 using the NheI site in one end and blunt end ligation in the other. The chimeric gene was then subcloned as a 2.9-kilobase pair EcoRI-XhoI fragment in the centromeric pFL37 plasmid, which was derived from pFL38 (44) as reported previously (45).
pFL37P0/1␤-A similar cloning strategy was used in this case, but the intron present in the P1␤ protein-encoding RPP1B gene (43) was removed beforehand by overlap extension polymerase chain reaction (46).

Ribosome Extraction
Yeasts were grown exponentially in rich YEP medium up to A 600 ϭ 0.6, and cells were collected by centrifugation and washed with buffer 1 (100 mM Tris-HCl, pH 7.4, 20 mM KCl, 12.5 mM MgCl 2 , 5 mM ␤-mercaptoethanol). Cells in buffer 1 were supplemented with protease inhibitors (0.5 mol of phenylmethylsulfonyl fluoride, 1.25 g of leupeptin, aprotinin, and pepstatin/g of cells) and broken with glass beads. The extract was centrifuged in a Beckman SS-34 rotor (12,000 rpm, 15 min, 4°C) yielding the supernatant S30 fraction, which was afterward submitted to high speed centrifugation at 90,000 rpm for 30 min at 4°C in a Beckman TL100.3 rotor. The supernatant S100 fraction was stored at Ϫ80°C, and the crude ribosome pellet was resuspended in buffer 2 (20 mM Tris-HCl, pH 7.4, 500 mM NH 4 Ac, 100 mM MgCl 2 , 5 mM ␤-mercaptoethanol). When required, ribosomes were centrifuged through a discontinuous sucrose gradient (20%, 40%) in buffer 2 at 90,000 rpm for 120 min at 4°C in a TL100.3 rotor. The pellet of washed ribosomes was dissolved in buffer 1 and stored at Ϫ20°C.

Protein Analysis
Ribosomal proteins were analyzed either by 15% SDS-polyacrylamide gel electrophoresis or by isoelectrofocusing. Isoelectrofocusing was carried out as described previously (47). Particles were pretreated with RNase A (10 g/mg of ribosomes) on ice for 30 -45 min. After lyophilization, the samples were resuspended in a loading buffer (6% ampholytes, 8 M urea) and directly loaded into a standard vertical gel (5% acrylamide, 0.2% bisacrylamide, 6 M urea, 6% pH 2.5-5.0 ampholytes). 30 mM NaOH and 180 mM H 2 SO 4 were used as cathode and anode solutions at the upper and lower part of the gel, respectively. Isoelectrofocusing was run in the cold room at 6-mA constant current until the voltage reached 600 V and then was lowered to 250 V for 16 h.
Proteins were usually detected by standard silver staining. Alternatively gels were stained in a solution containing 0.25% Coomassie R-250 (Sigma) dissolved in 45% ethanol, 10% acetic acid. After 30 min, the gel was destained using the same solution without stain.

Western Blotting
Proteins in gels were transferred to membranes, which were treated with 5% skimmed milk dissolved in TBS (10 mM Tris-HCl, pH 7.4, 200 mM NaCl) for 30 min, and afterward they were incubated for 1 h with the antibody diluted in the same buffer. Subsequently the membranes were washed for 15 min in TBS containing 5% skimmed milk and 0.1% Tween 20, and then the second antibody (R␣M/PO or D␣R/PO), diluted in the former buffer, was added and incubated for 30 min. Finally the membrane was washed for 15 min with 0.1% Tween 20 in TBS. Bound antibodies were located by detecting peroxidase activity using the ECL system (Amersham Pharmacia Biotech) and then exposed to film. Specific antibodies to the different P proteins were described previously (38,48).

Functionality of Protein P0 Chimeras Carrying Different
Acidic Proteins as Carboxyl-terminal Domain-Four protein P0 chimeras were constructed, P0-1␣, P0-1␤, P0-2␣, and P0-2␤ in which the last 112 residues in the P0 amino acid sequence were replaced by each one of the acidic proteins, P1␣, P1␤, P2␣, and P2␤ (Fig. 1). The overall amino acid sequence identity of the acidic proteins and the replaced P0 segment ranges from 31.5 to 35.4%, which is reduced to around 25% if the identical last 13 residues are excluded. In addition, an alanine-and glycine-rich region, equivalent to the hinge of the eukaryotic acidic proteins (24,37), is found at an equivalent position in the P0 fragment (Fig. 1). The chimeric genes, subcloned in the plasmids pFL37P0/1␣, pFL37P0/1␤, pFL37P0/2␣, and pFL37P0/2␤ were used to transform the S. cerevisiae P0 conditional null strains W303dGP0, D4567dGP0, D45dGP0, and D67dGP0. A summary of the strains and plasmids used is shown in Table I. In these strains, the genomic P0 gene was replaced by a gene copy under the control of the GAL1 promoter (38). Consequently the strains depend on the P0 gene in the transforming plasmid to be able to grow in glucose media. As a control, the strains were also transformed with pFL37P0, which contains a wild-type copy of the P0 gene. S. cerevisiae W303dGP0 contains the nuclear genes for the four acidic stalk proteins, whereas strain D4567dGP0 lacks the four genes. D67P0 and D45dGP0 lack the nuclear genes for both P1 (P1␣/P1␤) and both P2 (P2␣/P2␤) proteins, respectively (26,27).
All the transformed strains were able to grow on glucose agar plates as well as in liquid media (Table II), indicating that the gene chimeras can complement the absence of the wild-type P0 expression in a glucose medium. In D4567dGP0, complementation depends exclusively on the plasmid-encoded P0 protein because there are no acidic proteins expressed in this strain (27). As expected from previous reports (27), wild-type P0, expressed from the control pFL37P0, allowed cell growth in glucose. In this case, replacement of the P0 carboxyl-terminal domain by either protein P2␣ or P2␤ did not significantly affect the functionality of the resulting P0-2␣ and P0-2␤ proteins, whereas the expression of both P1 chimeras, either P0-1␣ or P0-1␤, caused a small but clear reduction of the protein-complementing activity. In contrast, the two P0/P1 chimeras were less damaging than the P0/P2 derivatives in S. cerevisiae W303dGP0. In the double disruptants D45dGP0 and D67dGP0, a stimulation of cell growth occurred when the chimera provided the acidic protein type missing in the corresponding strain. Thus, P0-1␣ and P0-1␤ stimulated growth in the D67 P1-defective cells, whereas P0-2␣ and P0-2␤ stimulated growth in the D45 strain lacking both P2 proteins.
Composition of the Ribosomal Stalk in Cells Expressing P0 Chimeras-Isoelectrofocusing showed that ribosomes from W303dGP0 transformed with the plasmids expressing the P0 chimeras were deprived of most of the 12-kDa proteins ( Fig.  2A). Only one of them was detected in each case. Thus, proteins P2␤, P2␣, P1␤, and P1␣ were found in cells expressing P0-1␣, P0-1␤, P0-2␣, and P0-2␤, respectively (Fig. 2). P1␤ was mainly present in a processed form called P1␤Ј (49). No traces of the remaining proteins were found in the washed ribosomes. In contrast, acidic proteins accumulated free in the S100 supernatant fraction of the cells expressing the chimeras (Fig. 2B).
A similar analysis of the P2-defective D45dGP0 transformants showed that ribosomes carrying P0-2␣ only bound protein P1␤, whereas those containing P0-2␤ exclusively bound P1␣. The ribosomes from the same strain transformed with the P0-1␣ and P0-1␤ constructs did not contain any acidic protein in the ribosome (Fig. 3A). In the case of the P1-defective D67dGP0-derived strains the isoelectrofocusing gels showed the presence of P2␤ and P2␣ in the ribosomes from cells carrying the P0-1␣ and P0-1␤ chimeras and the absence of acidic proteins in the particles having P0-2␣ and P0-2␤ (Fig. 3B).
Effect of P0 Chimeras in Growth at 37°C-Alterations in the ribosomal stalk have previously been shown to induce some specific phenotypes when grown at 37°C (47,50). In this study, when the W303dGP0 transformants were grown at this temperature, the functionality of all P0 chimeras was clearly reduced, whereas D45dGP0 and D67dGP0 were unable to grow even when expressing a wild-type P0. However, growth was rescued in all cases by the presence of a moderately high salt concentration, which has no effect at 30°C (Fig. 5).
An interesting singularity was the P0-1␤-transformed a Protein encoded in the genome of the yeast strain and expressed in the cell. The wild-type RPP0 gene is under the control of the GAL1 promoter in all cases (38).
b Protein chimeras encoded in the transforming plasmid. The fused genes are under the control of the native 5Ј and 3Ј untranslated region from the native RPP0 gene (see "Materials and Methods"). The numbers in the last column correspond to the total number of amino acids in each protein chimera. Excluding the last 13 amino acids, the overall amino acid identity of P0, estimated by the Align search program, is 25% for P1␣, 23% for P1␤, 25% for P2␣, and 26% for P2␤. D67dGP0 strain, which was able to grow at 37°C in the absence of salt. To test whether this peculiarity is because of a difference in the translating capacity, the activity of ribosomes from the temperature-sensitive D67dGP0/P1␣ and the temperature-insensitive D67dGP0/P1␤ was estimated. Ribosomes from cells grown at 30°C were tested in a poly(U)-dependent polymerization assay at 30°C and 37°C using as a control particles from wild-type W303 (Table III). As expected, all samples were less active at 37°C than at 30°C. The ribosomes from D67dGP0/P1␣ seemed to be a little more affected at 37°C than those from D67dGP0/P1␤. It is, however, unlikely that this difference in translation activity can explain the incapacity of D67dGP0/P1␣ to grow at 37°C in the absence of salt.
Temperature did not affect the stalk composition from strains D67dGP0/1␣ and D67dGP0/1␤. As in cells grown at 30°C, only protein P2␤ was found in the ribosomes containing P0-1␣ and only protein P2␣ was found in those with P0-1␤ when cells were placed at 37°C (Fig. 6). No protein was detected in the cells from D67dGP0 kept in the same conditions as was expected. Thus, the different growth displayed by these two strains at 37°C seems not to be due to an alteration of the stalk composition at this temperature. DISCUSSION The basic function of the stalk in the protein synthesis process is related to the activity of the supernatant factors. The bacterial stalk proteins, mainly L7/L12, interact with the factors in a very dynamic way, displaying different conformations during the elongation process (51,52). A similar situation seems to occur in eukaryotes (12).
In contrast to the bacterial stalk components L10 and L7/ L12, the carboxyl-end domain is highly conserved in the eukaryotic stalk proteins P0 and P1/P2 (13). In fact, the 100amino acid long carboxyl-terminal domain of P0 and the P1/P2 proteins plays a similar role. Because of this functional equivalence, the acidic proteins are not essential for ribosome function but modulate its activity (36). In the absence of P1/P2, the whole P0 protein is required for cell viability, and a truncation of its carboxyl end leaving only the amino-terminal domain totally inactivates the ribosome (25).

FIG. 2. Acidic 12-kDa proteins in ribosomes (A) and S100 fraction (B) from S. cerevisiae
W303dGP0 containing the four acidic proteins and expressing either the wild-type P0 (wt) or the P0 chimeras as indicated. A, ribosomes (0.5 mg) were resolved by polyacrylamide gel isoelectrofocusing in a 2.4 -5.0 pH range as indicated under "Materials and Methods." Proteins were detected by silver staining. The position of each acidic protein is indicated. B, total cell extracts from S. cerevisiae were centrifuged to remove ribosomes, and aliquots of the resulting S100 supernatant fraction were resolved by SDS-polyacrylamide gel electrophoresis. Proteins were detected by Western blot using monoclonal antibody 3BH5 specific to the carboxyl end of all P1/P2 proteins. P1/P2 marks the position of the 12-kDa protein band. domain of protein P0 can be replaced by any one of the four acidic 12-kDa proteins without notably affecting its activity in S. cerevisiae D4567, which contains ribosomes carrying only this stalk component. The last 12 amino acids, ESDDDMGF-GLFD, are identical in P0 and P1/P2 proteins, whereas the rest of the sequences show low amino acid identity (Fig. 1); this peptide is, therefore, the main common requirement for basic stalk activity probably because of its interaction with some conserved region in the elongation factors.
Although one carboxyl-terminal peptide, regardless of its origin, is enough to allow the stalk to perform its function, the efficiency of the translation machinery is related to the overall number of acidic proteins in the ribosomal stalk. Thus, when the P0 chimeras are expressed in D45dGP0 and D67dGP0,which express either the P1-type or the P2-type proteins, the growth rate of the strain increases because one P protein is able to bind to the ribosome, providing in this way two carboxyl domains to the stalk. On the contrary, the growth rate of W303dGP0, which contains the four acidic proteins, is diminished by the presence of the P0 chimeras, which cause a reduction of the 12-kDa proteins in the ribosome.
The stalk composition seems to have not only quantitative but also qualitative effects on the translation process (27). Most D45dGP0 and D67dGP0 transformants show a temperaturerelated osmosensitive phenotype. However, the different growth capacity of S. cerevisiae D67P0/1␤ and D67P0/1␣ at 37°C is significant. These data confirm that the qualitative composition of the stalk and not only the number of bound acidic proteins determine the capacity of the cell to grow in certain conditions. This differential growth seems not to be due to important differences in the overall efficiency of their respective translation machineries (Table III) but probably to the specific effect on the synthesis of some proteins by ribosomes carrying a particular stalk composition. Similar phenotypes have been found in other ribosomal stalk mutants (47,50). We are presently exploring whether or not a specific component in the cellular cell wall integrity signal transduction route is affected in these strains. So far no significant alterations in the dual phosphorylation of the mitogen-activated protein kinase Slt2 kinase, which is usually activated in cell wall-defective strains (53), have been detected in our mutants. 1 Other components of the same route are being tested in the hope of identifying proteins whose expression is affected by the ribosomal stalk composition.
The most interesting conclusions from this report are probably those concerning the overall stalk structure. The presence of four acidic proteins makes the yeast ribosomal stalk different from bacteria and even other eukaryotes, raising interesting structural questions with evident functional implications. An obvious issue is the interaction among the different stalk components. The bacterial counterpart, protein L7/L12, has been shown to be present as dimers in the ribosome (1) as well as P1 and P2 in some eukaryotes (23). However, the results indicate that it is not the case in S. cerevisiae. Thus, yeast ribosomes carrying a P0 chimera only bound one 12-kDa acidic protein, and this was never the one that was fused to the chimeric P0 (Fig. 5). The most direct conclusion from these data is that none of the yeast 12-kDa proteins is able to self-associate to form   6. Acidic proteins in ribosomes from S. cerevisiae D67dGP0, D67dGP0/1␣, and D67dGP0/1␤ grown at 37°C. Strains were grown at 30°C and then shifted to 37°C for 15 h. Ribosomes were resolved by SDS-polyacrylamide gel electrophoresis, and the proteins were detected by Western blot using specific monoclonal antibodies to either P2␤ (A) or P2␣ (B). As a positive control, ribosomes from the parental S. cerevisiae W303 were included (wt). dimers in the ribosome. They are present as monomers, indicating that contrary to the bacterial proteins they do not need to be dimerized to be functional. Nevertheless, the yeast acidic proteins do not function independently. Our results clearly show that there is a favored association between P1␣ with P2␤ on the one hand and P1␤ with P2␣ on the other. These two protein pairs might be considered as heterodimers playing the role that the two L7/L12 dimers do in bacteria.
Data pointing to a different structural and functional role for the P1␣/P2␤ and P1␤/P2␣ pairs were reported earlier. Thus, both protein couples are differentially affected by the absence of protein L12, which interacts with the GTPase rRNA near the P0-P1/P2 binding site (54). A similar situation was found when S. cerevisiae P0 was replaced by the equivalent protein from other species (45). Moreover, the two acidic couples have a differential effect on the resistance of the ribosome to some translocation inhibitors (55). Physicochemical studies have indicated an interaction between P1␣ and P2␤ in solution (56), and it was also recently reported that protein P2␣ but not P2␤ protects P1␤ from degradation in the cell, confirming the specificity of the P1␤/P2␣ association (57).
The available data indicate that the wild-type yeast ribosomal stalk is formed by the interaction with protein P0 of two acidic protein complexes, P1␣⅐P2␤ and P1␤⅐P2␣, which already seem to be formed in the cell cytoplasm (57). Protein P0 seems to be able to directly interact only with the P1 proteins as shown previously by ribosome reconstitution tests (58) and double hybrid data (36), and it is possible, therefore, that a direct interaction between P0 and the P2 proteins may not exist in the ribosome. There are at least two ways in which P1␣⅐P2␤ and P1␤⅐P2␣ can be assembled in the stalk (Fig. 7). In one of them the two proteins of the same type are adjacent at one of the ends. In the other, the four proteins alternate in such a way that the proteins of the same type can hardly touch because they are separated by a protein of the opposite type. There is not enough experimental information to decide which of the possible models is correct, and a detailed study of the interaction among all the different stalk components using different experimental approaches is being carried out in an attempt to resolve this question.