The yeast ribosomal GTPase associated center is made of parts of the 26S rRNA domains II and VI, and a number of proteins including P0, P1α, P1β, P2α, P2β and L12. Mapping of the rRNA neighborhood of the proteins was performed by footprinting in ribosomes from yeast strains lacking different GTPase components. The absence of protein P0 dramatically increases the sensitivity of the defective ribosome to degradation hampering the RNA footprinting. In ribosomes lacking the P1/P2 complex, protection of a number of nucleotides is detected around positions 840, 880, 1100, 1220–1280 and 1350 in domain II as well as in several positions in the domain VI α-sarcin region. The protection pattern resembles the one reported for the interaction of elongation factors in bacterial systems. The results exclude a direct interaction of these proteins with the rRNA and are compatible with an increase in the ribosome affinity for EF-2 in the absence of the acidic P proteins. Interestingly, a sordarin derivative inhibitor of EF-2 causes an opposite effect, increasing the reactivity in positions protected by the absence of P1/P2. Similarly, a deficiency in protein L12 exposes nucleotides G1235, G1242, A1262, A1269, A1270 and A1272 to chemical modification, thus situating the protein binding site in the most conserved part of the 26S rRNA, equivalent to the bacterial protein L11 binding site.
Received July 13, 2000; Revised and Accepted October 2, 2000.
Aminoacyl–tRNA binding to the A site and translocation of the A site bound peptidyl–tRNA to the P site after peptide bond formation require the interaction of two different G proteins, the elongation factors EF-Tu (EF-1a) and EF-G (EF-2), which bind to almost identical sites in the large ribosomal subunit [when the names of equivalent elements (elongation factors, ribosomal proteins nucleotide positions) in different organisms are given, the first corresponds to prokaryotes and the second, in parentheses, to eukaryotes]. The different ribosomal elements required for stimulation of the low intrinsic GTPase activity of both factors form the GTPase associated region of the ribosome, or GTPase center. At least two well-defined regions of the large rRNA form part of the GTPase center, the α-sarcin loop in domain VI and a T-shaped region from around nucleotides 1010 to 1130 in the secondary structure of the Escherichia coli 23s rRNA domain II. In addition, a number of ribosomal proteins have also been implicated in the GTPase center, including proteins L11 (L12), L10 (P0) and L7/L12 (P1/P2). Proteins L10 (P0) and L7/L12 (P1/P2) form a pentameric protein complex which constitutes one of the typical lateral protuberances of the large ribosomal subunit, the so-called ‘stalk’. The ‘stalk’ binds through the N-terminal domain of protein L10 (P0) to the vertical bar of the T-shaped GTPase center rRNA domain (1,2), while protein L11 (L12) interacts with the crossing bar (3). The complex of eubacterial protein L11 along with a 58 nt fragment has been crystallized and its 3-D structure resolved at 2.8 Å resolution (4,5).
The recent publication of a high resolution model of the prokaryotic large ribosomal subunit confirmed the position of most of the GTPase center components relative to the rest of the particle (6). Unfortunately, L7/L12 and the N-terminal domain of L11 did not clearly show up in the density map probably due to their mobility.
Considering the functional and structural conservation of the GTPase RNA domain and protein components, a similar structure is assumed for the eukaryotic rRNA–L12 complex (5). Nevertheless, variations in certain rRNA and protein regions must affect the eukaryotic structure, explaining the clear functional differences existing among kingdoms. These differences are especially noticeable in the structure and function of the stalk. A wealth of biochemical data indicated an important role for the bacterial stalk in the interaction and function of the elongation factors (7), which has been recently confirmed by cryo-electron microscopy (8,9) as well as genetically (10). Experimental evidence indicates a similar function for the eukaryotic stalk (11–14), but, in addition, the data are compatible with the involvement of this structure in a translation regulatory mechanism. Important structural features, which confer a very dynamic character to the eukaryotic stalk, support this new function (see 15 for a review on the eukaryotic stalk). Thus, in contrast to protein L7/L12, the eukaryotic acidic proteins are not essential for ribosome activity, and ribosomes totally deprived of P1/P2 proteins are functional but translate a partially different set of proteins (16). This important characteristic of the eukaryotic ribosome is caused by protein P0. This protein has a C-terminal extension, absent in the bacterial L10, which structurally and functionally resembles the P1/P2 proteins, and constitutes a minimal stalk that allows ribosome activity (17).
The L10 (P0) binding domain in the GTPase RNA is highly conserved among eukaryotes (18). Moreover, P0 binds and partially inactivates eubacterial ribosomes when the eukaryotic protein is expressed in E.coli cells (M.A.Rodriguez-Gabriel, M.Remacha and J.P.G.Ballesta, unpublished results). Similarly, protein L10, as part of the L10(L7/L12)4 pentameric complex, is able to functionally bind to rat rRNA (19). Nevertheless, the binding characteristics of the eukaryotic protein differ notably from those of the eubacterial polypeptide; for instance, P0 affinity for the RNA is higher, and the bound protein totally resists washing conditions that easily remove protein L10 from the bacterial ribosome (20).
Not surprisingly, the binding of protein L11 affects the binding of the L10(L7/L12)4 complex with the bacterial ribosome (21). A mutual but different interaction of these two sets of proteins also takes place in eukaryotes. Thus, in the absence of the eukaryotic protein L12, part of the P1/P2 protein is released from the ribosomes (22) while bacterial protein L11 does not affect the amount of bound L7/L12 (23).
All these data strongly indicate that the GTPase center, and especially the stalk, in spite of evident similarities, is different in the eubacterial and eukaryotic ribosomes, which may be reflected in the way the protein and RNA moieties interact. The archaebacterial stalk components seem to resemble the eukaryotic ones (24,25), but detailed functional and structural data are still missing. Footprinting is a classic way to explore ribosomal RNA–protein interactions (26), and this approach has provided important information in the case of the eubacterial GTPase center (1,27,28). The results from a recently reported footprinting study on rat liver using a GTPase center reconstituted in vitro from its RNA and protein components agree well with the bacterial data (29). The availability of a number of S.cerevisiae mutants lacking each one of the GTPase center protein components, (16,22,30), led us to perform a footprinting analysis in more physiological conditions in an attempt to study how the peculiar yeast stalk composition affects the RNA–protein interactions in this active center.
MATERIALS AND METHODS
Yeast strains and growth media
Saccharomyces cerevisiae strains W303 (MAT α, leu2-3, 112, ura3-1, trp1-1, his3-11, 15, ade2-1, can1-100), W303dGP0 (MAT α, leu2-3, 112, ura3-1, trp1-1, his3-11, 15, ade2-1, can1-100, rpP0::URA3-GAL1-rpP0) (30), D4567 [MAT α, leu2-3, 112, ura3-1, trp1-1, his3-11, 15, ade2-1, can1-100, rpY1α::LEU2, rpYP1β::TRP1, rpYP2α(L44)::URA3, rpYP2β(L45)::HIS3] (16) and 6EA1 (MAT α, leu2-3, 112, ura3-1, trp1-1, his3-11, 15, ade2-1, can1-100, rpL12A::KAN, rpL12B::HIS3) (22)were grown at 30°C in YEP medium (1% yeast extract, 2% peptone). The carbon source was either 2% glucose or 2% galactose as indicated. To obtain ribosomes deprived of protein P0, S.cerevisiae W303dGP0 growing in YEP–galactose was transferred to YEP–glucose medium for 9 h.
Preparation of ribosomes
Cells in 20 mM Tris–HCl pH 7.4, 80 mM KCl, 10 mM MgCl2, were broken with glass beads and the extracts were centrifuged for 15 min at 15 000 r.p.m. in a Sorvall SS34 rotor, producing an S-30 fraction which was centrifuged to obtain the ribosomes and supernatant fractions as previously described (14). When appropriate, ribosomes were washed twice by centrifugation through 20–40% sucrose in 10 mM Tris–HCl pH 7.4, 100 mM MgCl2, 500 mM NH4Cl, 5 mM β-mercaptoethanol. Ribosomes were finally resuspended in 10 mM Tris–HCl pH 7.4, 80 mM KCl, 12.5 mM MgCl2 and 5 mM β-mercaptoethanol and stored at –80°C.
Total rRNA preparation
Ribosomes (300 µg) were resuspended in 250 µl of 300 mM NaAc2 and 25 mM H3BO3. Total rRNA was phenol extracted by standard methods and renatured in 10 µl of TM4 buffer (10 mM Tris–HCl pH 7.4 and 4 mM MgAc2). The integrity of the rRNA was determined after denaturation of the samples at 70°C for 10 min by 3% polyacrylamide gel electrophoresis at 55°C and methylene blue staining (31).
Chemical reagents used for rRNA modification were DMS (di-methyl sulphate, Merck) for A(N-1) > C(N-3), kethoxal (2-keto-3-ethoxybutyraldehyde, United States Biochemical) for G(N-1,N-2) and CMCT [1-cyclohexyl-3-(2-morpholinoethyl)-carbodiimide metho-p-toluene-sulphonate, Sigma] for U(N-3) > G(N-1). These chemical reagents are single-strand RNA specific and monitor unpaired bases, but some RNA double helical irregularities can also be detected (26). Such altered stem positions include hydrogen-bonded adenines in the syn conformation (which can be modified by DMS) and G–C pairs at the end of the helices (sometimes reactive to kethoxal). DMS also reacts against guanosines at the N-7 position, although these modifications are not detected by reverse transcription.
Ribosomes (20 µg) were incubated at 30°C for 20 min in 100 µl of 50 mM HEPES–KOH pH 7.8, 15 mM KCl, 15 mM NH4Cl, 10 mM MgCl2, 1 mM DTT and 0.1 mM EDTA. As a control of the accessibility to the chemical reagents of non-complexed ribosomal RNA, 12 µg of renatured RNA were incubated in the same conditions in 70 mM HEPES–KOH pH 7.8, 270 mM KCl, 10 mM MgCl2 and 1 mM DTT. Controls of non-modified ribosomes and rRNA were also included.
Aliquots (100 µl) of ribosomes and naked rRNA were treated with each of the following chemical probes: 1 µl DMS (1:1 dilution in ethanol), 5 µl kethoxal (35 mg/ml in 20% ethanol) and 100 µl CMCT (42 mg/ml in the preincubation buffer). Samples were incubated at 30°C for either 10 min (DMS and kethoxal) or 20 min (CMCT). Detailed modification procedures, precipitation reactions and phenol extraction of modified rRNA from ribosomes are described by Christiansen et al. (26).
Reactive positions at domains II and VI of the 26S rRNA were identified by extension of 5′ 32P-labeled oligonucleotide primers using reverse transcriptase (Amersham Life Science) and by sequencing gel electrophoresis according to the method of Moazed et al. (32). The primers were complementary to the following sequence regions of the 26S rRNA: 821–837, 992–1008, 1159–1175, 1356–1372, 1463–1479 and 3112–3128. The approximate magnitudes of the altered nucleotide reactivities were estimated by microdensitometry and averaged over several (three to six) experiments after normalizing the signals with respect to bands representing non-specific stops (control cuts) to account for differences in the loading.
Binding of the antifungal GM193663A to the ribosome
The water-soluble sordarin derivative GM193663A (Glaxo Wellcome) (33) was added to an S-30 fraction at a final concentration of 0.1 mM, and incubated at 30°C for 30 min. The mixture was then cooled and placed on ice (26). Ribosomes were obtained as previously indicated. In order to preserve the interaction of the inhibitor with both ribosomes and elongation factors, the antibiotic was present at the same concentration in the buffers in all preparation steps.
RESULTS AND DISCUSSION
The set of oligonucleotides used for primer extension enabled us to scan the complete domain II (positions 644–1457) and the neighborhood of the α-sarcin loop in domain VI (positions 2840–3107) of S.cerevisiae 26S rRNA after chemical modification. Footprinting was performed on ribosomes lacking different stalk ribosomal proteins using intact particles as well as naked 26S rRNA as a control. Specific conformational changes due to the presence of GTPase center proteins were found, but first, some general structural characteristics of the scanned regions will be discussed.
Structural features of the S.cerevisiae 26S rRNA domain II and the α-sarcin region
Primer extensions on naked rRNA usually show a number of reverse transcriptase stops that occurred in all samples. These non-specific stops or ‘control cuts’ correspond to highly stable RNA secondary or tertiary structural features that are maintained at the reverse transcription temperature (26). Only two strong reverse transcriptase stops, totally blocking the reverse transcriptase activity, have been detected in the tested yeast rRNA regions, corresponding to positions A644 and C2840 (data not shown). Similar stops in eubacterial rRNA have been previously described as being indicative of a base-methylated residue 5′ to the stop. A strong stop at an A644 equivalent position has also been detected in phylogenetically distant organisms such as Drosophila melanogaster (34) and the halophilic archaeon Haloferax mediterranei (C.Briones and R.Amils, unpublished results). These results suggest that methylation at these positions seems to be a conserved feature of the large rRNA.
Comparison of results from chemical modification of naked rRNA and ribosomes showed, as expected, that nucleotides participating in helical regions are protected in both instances. However, stem regions, such as nucleotides 750–781 and 962–966 (Fig. 1A and B), which are modified in the naked RNA and protected in the ribosome, have also been detected. These double-stranded regions may not be formed in naked RNA, their Watson–Crick interactions being stabilized in the ribosome (see Fig. 2, for a model of the secondary structure of the studied 26S rRNA regions).
Almost all of the unpaired nucleotides are reactive in free 26S rRNA but not in the S.cerevisiae W303 ribosomal particles (Fig. 2). Protein–RNA interactions as well as some RNA tertiary structure features protect most of the loop regions from attack by the chemical probes or RNases, as in bacteria (35) and mammals (36). In fact, some nucleotides in unpaired regions, like those in the neighborhood of nucleotides 1173–1174 and 1310–1311, correspond to positions predicted to take part in tertiary interactions in eubacteria (37,38), suggesting that those long-range interactions also occur in yeast.
A number of nucleotides were found to be more reactive in ribosomes than in naked rRNA. Similar unprotected residues have been also described in E.coli ribosomes (35). They may reflect conformational changes associated with the assembly of the ribosomal particle, or may lie on functional sites of the ribosome, which require unpaired nucleotides to allow the interaction of ligands. The most prominent cases of this type in yeast 26S rRNA domain II are located at A689, A690, A691 (Fig. 1C), at C716 and A760 in the loop of the variable region V4 (Fig. 1A and C), at A1224 (Fig. 3) and at G1348–A1354 (Fig. 1D). Moderate stimulations were found in other positions shown in Figure 2.
Effect of stalk proteins on GTPase RNA reactivity
The 1218–1224 loop shows highly accessible and neighboring inaccessible nucleotides in S.cerevisiae ribosomes, following a pattern that was also found in ribosomes from E.coli (35). As previously commented, this region corresponds to the binding site of the S.cerevisiae P0 and associated acidic proteins. The accessibility changes observed in this loop upon protein binding may reflect a way to overexpose some nucleotides involved in functional interactions with translational ligands. The study of the effect caused on the nucleotide reactivity by specific proteins can provide information on their role in this process.
Using gene disruption methods, a set of S.cerevisiae mutants were obtained that lack one or more of the proteins involved in the ribosomal GTPase activity. The results from a footprinting study of the ribosomes from these mutants, summarized in Figure 2, have yielded information on the RNA regions directly or indirectly affected by the binding of these proteins.
Acidic protein deficient ribosomes. Saccharomyces cerevisiae D4567 was deprived of the four stalk acidic proteins, P1α,P1β, P2α and P2β, by gene disruption. The strain is viable in standard conditions but its functional ribosomes, which contain standard amounts of the remaining GTPase proteins, totally lack acidic proteins (16).
When the reactivity of the rRNA in ribosomes from wild-type S.cerevisiae W303 and D4567 strains was compared, a number of positions exhibited reproducible accessibility changes. All of them corresponded to a weak reduction of chemical reactivity in the mutant ribosomes. Most of these nucleotides were concentrated in the unpaired regions of the GTPase domain: A1224, G1241, G1242, A1243, A1244, A1269, A1270 and A1272 (Fig. 3A). Differential signals were also detected in the nucleotides A845, A846, A881 (Fig. 4A), A1078, A1092, A1101 and A1102 (Fig. 4B) as well as in the α-sarcin region (G3019, A3024 and G3025) (Fig. 4C). Most of these nucleotides were unreactive in the naked rRNA and became exposed in the wild-type ribosomes, but they were protected again in ribosomes lacking P1/P2 proteins. However, some of them, such as G1214/2, A1092, A1102 and G3025, which were already exposed in the rRNA, were protected in the defective ribosome as well.
These results seem to exclude a direct interaction of the P1/P2 proteins with the rRNA, since in this case the protein removal would be expected to result in an increase of the nucleotide reactivity. They are, however, compatible with an indirect effect on the rRNA conformation induced by the acidic protein binding. Alternatively, since most of the equivalent positions have been shown to be protected by the elongation factors in bacterial ribosomes (39), and some of them are also protected by EF-2 in mammals (40), these results might reflect a higher amount of supernatant factors bound to the mutant ribosomes. Supporting this possibility, the differences tend to be reduced by high salt washing, after which the mutant ribosome reactivity is closer to that of the wild-type particles. Since both ribosome preparations are processed and washed in the same way, these results would indicate a higher affinity of the mutant ribosomes for the elongation factors. If so, the presence of the 12 kDa acidic proteins would promote the exchange of the factors in the ribosome rather than facilitate their binding, as has been assumed so far.
A higher affinity of the particles for the supernatant factors due to the absence of the 12 kDa proteins could also provide an explanation for the lower translation efficiency of the mutant particles (16).
P0 protein deficient ribosomes.Protein P0 is essential for ribosome activity and cell viability, but P0-deficient particles can be obtained from a conditional P0 null mutant grown in restrictive conditions for a controlled period of time (30). These defective particles were used in an attempt to study P0-dependent conformational changes in the GTPase RNA. Unexpectedly, no signal was detected in the footprinting gels when different preparations of these particles were used in primer extension assays (data not shown). Primers designed to anneal into domains I, III, IV and V in the 26S rRNA showed the same negative results. Since the rRNA extracted from the particles prior to treatment did not show substantial differences with the controls these results seem to indicate that RNA degradation must take place during the chemical modification reaction. However, similar negative results were observed when the modification reaction was carried out at either 0°C (26) or in the presence of diethyl pyrocarbonate to minimize the activity of contaminating RNases.
The apparent high sensitivity of these ribosomes to degradation is exclusively due to the absence of protein P0 since the parental strain, S.cerevisiae W303 and other mutants derived from the same strain, like S.cerevisiae D4567 mentioned formerly, did not present any footprinting problems. It is, therefore, possible that the absence of protein P0 notably increases the susceptibility of rRNA to residual RNase activities in the ribosome preparation. P0 is an essential protein for the GTPase center and its absence could expose a remarkably RNase-sensitive domain in RNA structure.
Protein L12 deficient ribosomes. Protein L12 has an important role in the structure of the yeast GTPase center, binding to loops 1239–1247 and 1267–1272, which are accessible to chemical probes both in naked rRNA and in ribosomes in agreement with data on E.coli and H.mediterranei (35; C.Briones and R.Amils, unpublished results). These two loops fold over close together in the tertiary structural model of this part of the GTPase domain along with protein L11(L12) (4,5) and contribute to the binding site for antibiotics thiostrepton and micrococcin (41), elongation factors (42) and rat anti-28S RNA antibody (29). The bacterial L11 protein is located at the base of the stalk, and in yeast the equivalent protein L12 seems to affect the binding of the P0/P1/P2 complex since some of the stalk components are not found in the ribosome in its absence (22).
In order to characterize conformational changes caused by L12 in the RNA, the differential rRNA reactivity patterns of wild-type ribosomes (S.cerevisiae W303) and particles lacking protein L12 (S.cerevisiae 6EA1) were studied. Two strong protections at G1235 and A1262 and four weaker ones at G1242, A1269, A1270 and A1272 were detected in the presence of protein L12 (Fig. 3B). G1235 and A1262, corresponding to E.coli U1061 and A1088, are located in the region protected by the binding of L11 in bacteria (1,2,4,5). However, while A1088(A1262) forms a conserved long-range base pair with U1060 and interacts directly with protein L11, U1061(G1235) bulges out the helix and seems not to be involved in the protein interaction (Fig. 5). Interestingly, nucleotide A1899 in rat 28S rRNA, equivalent to yeast A1262, was also the most protected position upon in vitro binding of purified rat L12 to naked rRNA. In contrast, the reactivity of the rat equivalent to yeast G1235, nucleotide G1873, was not affected. In its place, A1887 became protected by rat L12 (29), while the reactivity of the equivalent nucleotide in yeast, A1251, was totally unmodified (Fig. 3B). These results confirm that the differences between the yeast and rat GTPase center are not limited to the number of different acidic proteins in the stalk, four in the first case and two in the second, but probably extend to structure of the RNA–protein complex. Alternatively, these differences might be due to the fact that in our case native L12-deficient ribosomes were used while an in vitro reconstituted protein–RNA complex was utilized in the rat system.
The nucleotides showing smaller reactivity changes, G1242, A1269, A1270 and A1272, lie on the terminal loops of the GTPase domain. These two loops come close in the 3-D structure of the L11–RNA complex, but they are not directly involved in the interaction with the protein, although its N-terminal domain is not far from them (5). In fact, they form together with an L11 proline-rich loop, the binding site of thiostrepton and microccocin, two classic inhibitors of the bacterial GTPase center (43). Confirming the proximity of L11, protection in both loops by the protein has been reported (1,2), and recently they have been targeted by direct hydroxyl probing using Fe(II) tethered to position 19 in L11 (28).
Apart from these six nucleotides, the full scanning of the domain II and the α-sarcin region revealed four additional weaker protections in positions A845, A846, A881 and A882 (Fig. 4A), which coincides with the opposite effect caused by the absence of the acidic P1/P2 proteins as discussed previously. These results suggest that this domain II region lies in the vicinity of the GTPase domain in the 3-D structure of the ribosome and, therefore, collaborates in the GTP-associated reactions involved in the translational process. This implication has not yet been reported in eubacteria, although proximity between both regions of domain II has been proposed based on data from H.mediterranei ribosome reconstitution experiments (C.Briones and R.Amils, unpublished data). Finally, the region comprising these positions was described as functionally active, being involved in tRNA translocation (44).
Taken together, the results confirm the similarity of the L11/L12–RNA complex structure in the different species previously suggested by biochemical data (45) but at the same time indicate the existence of structural peculiarities in the yeast ribosome that will require the 3-D structure to be established in order to be fully understood.
Effect of GM193663A, a yeast GTPase inhibitor sordarin derivative
The antifungal sordarin derivative GM193663A, has been shown to specifically inhibit yeast protein synthesis by interfering with the interaction of the elongation factor EF-2 with the ribosome (33,46). Resistance mutations to sordarin have been found in EF-2 as well as in the stalk protein P0 (10,47). A disturbance of the GTPase center by the binding of the inhibitor should, therefore, be expected. To check this possible effect, comparison of rRNA reactivity to chemical probes in S.cerevisiae W303 ribosomes obtained in the presence and in the absence of the antibiotic was carried out. An increase of the reactivity was detected in nine positions in the GTPase RNA from ribosomes carrying the inhibitor. The strongest stimulation was found in position G1241; a weaker effect was detected in A1224, A1243, A1244, A1269, A1270 and A1272 (Fig. 3C). Two additional signals were found at the α-sarcin loop in G3019 and G3025 (Fig. 4C).
As indicated previously, G1241 and A1269 are equivalent to the E.coli A1067 and A1095. These nucleotides, together with protein L11, are directly involved in the binding of the thiopeptide antibiotics, thiostrepton and microccocin (43). Like sordarin derivatives in yeast, these antibiotics block the function of the bacterial elongation factor EF-G. It may be tempting to conclude from our footprinting results that GM193663A acts in the yeast ribosome in a similar way to thiopeptide antibiotics in eubacteria, especially considering that microccocin, like the sordarin derivatives, stimulates the reactivity of A1067 (27). In fact, although the available information indicates that sordarin derivatives interact with the elongation factor (48), the resistance mutations found in protein P0 (10,47) indicate that the stalk structure must also be affected exposing some nucleotides, as shown in this report. Moreover, as previously commented on, these nucleotides are among those involved in elongation factor interaction, in agreement with the inhibition of EF-2 function by the sordarin derivatives. Interestingly, the EF-G inhibitor fusidic acid has an opposite effect to sordarins, protecting rather than exposing equivalent nucleotides in eubacterial ribosomes. It seems, therefore, that both types of compound act in a different way in agreement with the biochemical data (49).
The pattern of the nucleotide chemical accessibility in the GTPase associated center tested in the naked RNA and in the ribosomes from S.cerevisiae is not very different from the one reported for other systems (35), indicating a similar overall conformation. However, the notable differences existing in the protein moiety of this ribosomal active center in different organisms (15) was expected to bring about specific features in the yeast rRNA structure. This has been confirmed, and the most notable peculiarity is the different nucleotide reactivity caused by the absence of protein L12 in yeast and in rat. These data suggest that the differences in the GTPase center of mammals and yeast are not limited to the proteins but extend also to the RNA conformation. Comparative functional data on eukaryotic GTPase from different species are scarce, but they indicate significant differences. Probably the most interesting is the existence in mammals of a cytoplasmic pool of a P0/P1/P2 protein complex (50), which is missing in yeast (30,51). This might be related to different functional roles for this structure in higher and lower eukaryotes (see 15 for a discussion). Further structural data are obviously required to confirm this hypothesis.
Also worth noting is the apparent higher affinity of ribosomes lacking the P1/P2 protein set for the elongation factors. If this is so, the 12 kDa acidic proteins could somehow participate in the exchange of the factors during the elongation process rather than merely facilitate their binding to the ribosome, as is usually assumed. A kinetic study of the EF-2–ribosome interaction with different defective ribosomes is underway to test this possibility. The stalk acidic proteins could have a role in the proposed translocation mechanism in which the elongation factor EF-G(EF-2) serves as a motor protein using the free energy of GTP hydrolysis (52). In fact, several theoretical models for the involvement of the stalk proteins in the translocation process have been proposed (53,54).
The high sensitivity to degradation of ribosomes lacking protein P0 is probably the consequence of exposing a high RNase sensitive site. It is possible that the absence of other ribosomal RNA-binding proteins has similar effects. However, P0 is probably the only protein among them, at least in mammals, that can be exchanged and can be found free in the cytoplasm (50). This implies that at some moment in its life cycle the ribosome may lack P0, and that the RNase sensitive site is accessible to degrading enzymes. Could this be the first step of a ribosome degradation pathway?
Our results are compatible with an effect of sordarin derivatives on the elongation factor EF-2 interaction with the ribosomes (46,55), but, in addition, they clearly show a simultaneous distortion of the ribosomal stalk, exposing nucleotides related to the EF-2 interaction. These results indicate that sordarin derivatives and fusidic acid inhibit elongation factor function in a different way, confirming previous biochemical evidence (49). However, if these compounds inhibit translation in this way, it is surprising that mutations in ribosomal protein P0 can induce resistance to the drug without affecting notably the capacity of the drug to fix the factor to the ribosome (10,47). A reduction in the EF-2/sordarin/ribosome complex stability due to the P0 mutation has been proposed, but this proposal is difficult to reconcile with the results showing a similar sordarin binding Kd for wild-type and mutant ribosomes (47).
We thank M. C. Fernandez Moyano for expert technical assistance. This work was supported partially by grant PB94-0032 from the Dirección General de Política Científica (Spain), and by an institutional grant to the Centro de Biología Molecular from the Fundación Ramón Areces (Madrid).
To whom correspondence should be addressed. Tel: +34 91 397 5076; Fax: +34 91 397 4799; Email: email@example.com Present address: Carlos Briones, Centro de Astrobiología, INTA-CSIC, Carretera de Ajalvir km 4, 28850 Torrejón de Ardoz, Madrid, Spain