Replacement of L7/L12.L10 Protein Complex in Escherichia coli Ribosomes with the Eukaryotic Counterpart Changes the Specificity of Elongation Factor Binding*

The L8 protein complex consisting of L7/L12 and L10 in Escherichia coli ribosomes is assembled on the conserved region of 23 S rRNA termed the GTPase-associated domain. We replaced the L8 complex in E. coli 50 S subunits with the rat counterpart P protein complex consisting of P1, P2, and P0. The L8 complex was removed from the ribosome with 50% ethanol, 10 mm MgCl2, 0.5 m NH4Cl, at 30 °C, and the rat P complex bound to the core particle. Binding of the P complex to the core was prevented by addition of RNA fragment covering the GTPase-associated domain of E. coli 23 S rRNA to which rat P complex bound strongly, suggesting a direct role of the RNA domain in this incorporation. The resultant hybrid ribosomes showed eukaryotic translocase elongation factor (EF)-2-dependent, but not prokaryotic EF-G-dependent, GTPase activity comparable with rat 80 S ribosomes. The EF-2-dependent activity was dependent upon the P complex binding and was inhibited by the antibiotic thiostrepton, a ligand for a portion of the GTPase-associated domain of prokaryotic ribosomes. This hybrid system clearly shows significance of binding of the P complex to the GTPase-associated RNA domain for interaction of EF-2 with the ribosome. The results also suggest that E. coli 23 S rRNA participates in the eukaryotic translocase-dependent GTPase activity in the hybrid system.

Binding of translocases, EF 1 -G⅐GTP in prokaryotes and EF-2⅐GTP in eukaryotes, to a specific site of the ribosome causes GTP hydrolysis that drives translocation of peptidyl-tRNA from the A-site to the P-site during protein biosynthesis (1)(2)(3). The binding site of the prokaryotic EF-G has been identified on the universally conserved regions, the "GTPase-associated domain" surrounding residue 1067 and the "sarcin/ricin loop" of residues 2653-2667 in 23 S rRNA (4,5). Interaction of the eukaryotic EF-2 with the equivalent domains of 28 S rRNA has been also suggested (6 -8). Despite the highly conserved features of the two RNA domains which interact with the translocases EF-G and EF-2, these factors are not interchangeable between prokaryotic and eukaryotic translational systems.
Another component, long implicated in the ribosomal interaction with the translocase, is the stalk protein L7/L12 in prokaryotic ribosomes (9 -11). This protein together with L10 forms a stable pentameric complex, (L7/L12)2(L7/L12)2L10, termed L8 (12), and this complex binds to the GTPase-associated domain of 23 S rRNA through its interaction with L10 (13,14). The eukaryotic counterpart of the prokaryotic L8 is the "P complex" consisting of homodimers of P1 and P2 and monomeric P0 (15,16). Involvement of these constituent proteins in the activity of eukaryotic EF-2 has been demonstrated in vitro (17,18). An essential role of protein P0 for assembly of the complex into yeast ribosomes and for cell viability has also been shown in vivo (19,20). Rat P complex reconstituted from isolated P1, P2, and P0 specifically binds to the GTPase-associated domain of 28 S rRNA, probably through P0 protein (21,22). This binding site is nearly equivalent to that for the Escherichia coli L8 complex in 23 S rRNA, as determined by footprinting (23,24). In contrast with this similar RNA recognition feature of the stalk protein complexes, amino acid sequence homology of each protein constituent is very low between prokaryotic and eukaryotic counterparts (25,26). It is therefore presumed that the evolutionary divergence between prokaryotic L8 complex and eukaryotic P complex may parallel the luck of sequence similarity at the regions of EF-G and EF-2 interacting with their respective pentameric complexes.
We here attempt to replace L8 complex in E. coli ribosomes with the rat counterpart. The L8 complex is specifically detached from the GTPase-associated RNA domain within the ribosome in ethanol/NH 4 Cl (12,27), and subsequently the rat P complex is incorporated into the core ribosome. This replacement of the stalk protein complex changes the specificity of translocase interaction from prokaryotic EF-G to eukaryotic EF-2. The results demonstrate the functional significance of the ribosomal stalk protein complex and its binding to the GTPase-associated RNA domain for the ribosome-translocase interaction. The results also provide information about conserved features of rRNA involved in the translocase binding.

MATERIALS AND METHODS
Ribosomes and Core Ribosomes Lacking L8 Complex-E. coli Q13 cells (10 g) were ground with 20 g of alumina, and ribosomes were extracted with 20 ml of buffer A containing 10 mM MgCl 2 , 20 mM NH 4 Cl, 5 mM 2-mercaptoethanol, and 20 mM Tris-HCl, pH 7.5. The lysate was centrifuged twice for 30 min at 20,000 ϫ g. The supernatant was layered on 1.2 M sucrose cushion in Buffer A and ultracentrifuged for 14 h at 45,000 rpm (160,000 ϫ g) in a Hitachi P50A2 rotor at 4°C. The ribosome pellet was suspended in 20 ml of Buffer B containing 10 mM MgCl 2 , 0.5 M NH 4 Cl, 5 mM 2-mercaptoethanol, and 20 mM Tris-HCl, pH 7.5. Ribosomes were pelleted by ultracentrifugation for 3 h at 45,000 rpm with the same rotor. The salt wash was repeated two more times, and the ribosome pellet was resuspended in buffer A and stored at Ϫ80°C.
The L8 complex in E. coli 70 S ribosomes was removed according to * This work was supported by grants-in-aid for scientific research (10174212) and for COE Research (10CE2003) from the Ministry of Education, Science, Sports and Culture of Japan, and a fund from The Japan Society for the Promotion of Science (JSPS-RFTF96100305). 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.
‡ To whom correspondence should be addressed. Fax: ϩ81-268-21-5571; E-mail: uchiumi@giptc.shinshu-u.ac.jp. 1 The abbreviations used are: EF, elongation factor; P complex, ribosomal P protein complex consisting of P0, P1, and P2. Tris-HCl, pH 7.5, were preincubated at 30°C for 5 min. The solution was mixed with 1 ml of ethanol (prewarmed at 30°C) with stirring at 30°C. After 10 min, another 1 ml of ethanol was added, and stirring was continued for 5 min at 30°C. The solution was centrifuged at 25,000 ϫ g for 10 min, and the ribosomal pellet was dissolved in 2 ml of the extraction buffer. The same extraction was repeated once. Isolation of the resulting core ribosomes and split proteins was as described (27). By this treatment, proteins L7/L12 and L10 constituting L8 complex were predominantly released from the ribosomes, as observed by 16.5% SDS-polyacrylamide gel electrophoresis (28) (Fig. 1) and two-dimensional gel electrophoresis (29) (data not shown). Besides major proteins L7/L12 and L10, trace amounts of L1, L11, and S2 were also released from 70 S ribosomes by the extraction.
Preparation of Rat P Complex-The mammalian ribosomal P complex was reconstituted in vitro by mixing proteins P0, P1, and P2 isolated from rat liver ribosomes, as described previously (21). Formation of the complex was confirmed by 6% polyacrylamide gel electrophoresis under nondenaturing conditions (22).
Plasmid Construction and in Vitro RNA Synthesis-The DNA fragment comprising residues 1029 -1127 (GTPase-associated domain) of 23 S rRNA was amplified using the polymerase chain reaction (30) and inserted into HindIII and XbaI sites of an expression vector, pSPT18 (Roche Molecular Biochemicals). The RNA fragment was synthesized with SP-6 RNA polymerase and isolated, as described previously (31).
Binding of the P Complex to E. coli Core Ribosomes ("Hybrid" Formation)-The E. coli core 70 S ribosomes (130 pmol) were incubated with 16 g of the P complex in 1 ml of 15 mM MgCl 2 , 50 mM NH 4 Cl, 20 mM Tris-HCl, pH 7.5, at 37°C for 5 min. The sample solution was layered on 34 ml of 10 -28% linear sucrose gradient in 15 mM MgCl 2 , 20 mM NH 4 Cl, 20 mM Tris-HCl, pH 7.5. The gradient was centrifuged for 7 h at 23,000 rpm (68,000 ϫ g in the middle of the gradient tube) and 4°C in a Hitachi P28S swing rotor, and 1.2-ml fractions were collected from the bottom of the gradient. The ribosomes were located by absorbance at 260 nm. A portion of each fraction was precipitated with trichloroacetic acid and analyzed by SDS gel electrophoresis, followed by immunoblotting using anti-P monoclonal antibody which reacts with all P0, P1, and P2 (18).
Translocase-dependent GTPase Activity-The E. coli core ribosomes (2.5 pmol) were incubated with or without the rat P complex, as indicated in the figure legends, in 10 l of a buffer containing 5 mM MgCl 2 , 50 mM NH 4 Cl, 20 mM Tris-HCl, pH 7.5, at 37°C for 5 min. The eukaryotic EF-2-dependent GTPase hydrolysis was started by mixing another 10 l of the same buffer containing 0.5 g of EF-2, and 3 nmol [␥-32 P]GTP (60 -100 cpm/pmol). In the case of the prokaryotic EF-G-dependent GTP hydrolysis, the reaction was started by mixing the same buffer solution containing 0.5 g of EF-G, and 30 nmol of [␥-32 P]GTP (6 -10 cpm/pmol). To expand lower activities dependent on EF-2, we used 3 nmol of [␥-32 P]GTP with higher specific radioactivity instead of 30 nmol used for the prokaryotic EF-G-dependent activities. No marked difference of EF-2-dependent activity of rat 80 S ribosomes or hybrid ribosomes was observed between the two concentrations of GTP used (not shown). The reaction proceeded for 10 min at The Translocase EF-2 and EF-G-The mammalian translocase EF-2 was purified from pig liver, as descried previously (33). E. coli EF-G was purified as described by Kaziro et al. (32).
Rat Liver Ribosomes-Puromycin-and high KCl-treated 80 S ribosomes were prepared from rat liver, as described previously (34).

RESULTS
The eukaryotic ribosomal P complex is believed to be the counterpart of prokaryotic L8 complex that constitutes the ribosomal stalk. We have reconstituted the P complex from isolated rat proteins P0, P1, and P2 and shown its binding site in 28 S rRNA to be nearly equivalent to that for E. coli L8 complex in 23 S rRNA (22). We tested cross-binding of the rat P complex to the E. coli GTPase-associated domain by gel retardation using an RNA fragment comprising residues 1029 -1127 ( Fig.2A). The P complex strongly bound to the E. coli RNA domain (Fig. 2B, lane 3). Interestingly, the stability of this RNA-protein complex was higher than that with the cognate protein L8 (lane 2). This binding ability of the P complex for E. coli RNA was comparable with that for rat RNA (see Ref. 22).
To test whether the rat P complex binds to E. coli core ribosomes lacking the L8 complex (Fig. 1), the core ribosomes were incubated with the P complex and then analyzed by sucrose density gradient centrifugation and immunoblotting with anti-P monoclonal antibody reactive with P0, P1, and P2. The P complex bound to 70 S ribosomes and 50 S subunits, but not to 30 S subunits (Fig. 3A). These bindings were prevented by adding an excess amount of the RNA fragment encompassing the GTPase-associated domain (Fig. 3B). The results suggest that the rat P complex was incorporated into the E. coli ribosome through its binding to the GTPase-associated domain. We termed the resultant particles "E. coli-rat hybrid ribosomes." In this experiment, a significant portion of the core 70 S ribosomes was dissociated into 50 and 30 S subunits by ultracentrifugation in 15 mM MgCl 2 (upper panels), a condition that normally gives only 70 S couples. This dissociation occurred, irrespective of binding of P complex (see upper panels of Fig. 3, A and B), suggesting that removal of L8 complex from the 50 S subunits lowers their affinity for 30 S subunits.
The hybrid ribosomes as well as E. coli 70 S and rat 80 S ribosomes were tested for interaction with the translocase proteins by measuring their GTPase activity dependent on prokaryotic EF-G (Fig. 4A) and eukaryotic EF-2 (Fig. 4B). Intact E. coli (70S) and rat ribosomes (80S) had activities specific to prokaryotic EF-G and eukaryotic EF-2, respectively. The activity of the former was about 10-fold higher than the latter. The hybrid ribosomes (70S-P) showed eukaryotic EF-2-dependent activity ( Fig. 4B) but not prokaryotic EF-G-dependent activity (Fig. 4A). The level of EF-2-dependent activity of the hybrid ribosome was comparable with that of rat 80 S ribosomes ( Fig.  4B) but lower than that of intact E. coli ribosomes dependent on EF-G (Fig. 4A). The induction of EF-2-dependent activity of E. coli core ribosomes was apparently because of addition of the P complex. The P complex alone, however, was inactive (Fig. 5).
Induction of EF-2-dependent activity by mixing the E. coli core ribosomes and rat P complex was prevented by adding the RNA fragment encompassing the GTPase-associated domain, a competitor for the binding site of P complex (Fig. 6). This result is consistent with a failure of the hybrid formation by adding the RNA competitor (Fig. 3B) and suggests that binding of the P complex to the GTPase-associated domain is crucial for EF-2-dependent GTPase activity.
The antibiotic thiostrepton binds to the GTPase-associated domain of E. coli ribosomes and inhibits GTPase activity dependent on prokaryotic EF-G (35). This drug was used to test whether the RNA domain of E. coli ribosomes participates in EF-2-dependent GTPase activity in the hybrid ribosome. As shown in Fig. 7, the EF-2-dependent activity of the hybrid ribosome, unlike that of the rat 80 S ribosome, was inhibited by thiostrepton, although its inhibition efficiency was not as high as for the intact E. coli ribosome. This result suggests that the same RNA domain of E. coli 23 S rRNA is involved in the eukaryotic EF-2-dependent GTPase activity of the hybrid ribosome as well as in the prokaryotic EF-G dependent activity of the native 70 S ribosome.

DISCUSSION
It is generally recognized that the eukaryotic translocase EF-2 does not function with the prokaryotic 70 S ribosome, and the prokaryotic factor EF-G is not accessible to the eukaryotic 80 S ribosome (36,37). The present study demonstrates that replacement of the stalk protein complex L10-L7/L12 (L8 complex) in E. coli ribosome with rat counterpart P0-P1-P2 (P complex) changes the factor binding specificity from EF-G (homologous) to EF-2 (heterologous). This hybrid system provides clear evidence that the stalk protein complex is the major ribosomal component responsible for kingdom-dependent specificity of the ribosome-translocase interaction. We have also attempted the replacement of protein L7/L12 component of E. coli ribosomes with the rat counterpart P1/P2 by removing specifically L7/L12 from the ribosomes (27) and mixing the core with the rat proteins, but we failed to detect neither activity dependent on prokaryotic EF-G nor eukaryotic EF-2 (data not shown). Our results imply that whole bodies of the stalk protein complexes, L10-L7/L12 in prokaryotes and P0-P1-P2 in eukaryotes are the functional units which can be replaced by each other. This may be confirmed by replacement of P complex in the eukaryotic ribosomes with prokaryotic L8 complex, although it is technically difficult up to now.
Rat P complex apparently binds to the E. coli core ribosome by its direct interaction with the GTPase-associated RNA domain of 23 S rRNA, presumably through P0 moiety. This is supported by evidence that the P complex strongly cross-binds to the RNA fragment covering E. coli GTPase-associated domain (Fig. 2) and that binding of the P complex to the core ribosome is prevented by addition of the RNA fragment as a binding competitor (Fig. 3). This binding causes induction of eukaryotic EF-2-dependent GTPase activity of the ribosome (Figs. 4 -6). These data show the functional significance of the RNA-protein interaction. Moreover, it is notable that the induced EF-2-dependent GTPase activity is inhibited by the antibiotic thiostrepton which specifically binds to the prokaryotic GTPase domain and inhibits the prokaryotic EF-G-dependent GTPase activity (Fig. 7). This implies that the E. coli GTPase RNA domain has an ability to participate in the eukaryotic EF-2-dependent GTPase activity in the presence of the P complex.
It is noteworthy that the hybrid ribosome shows some resistance to inhibition by thiostrepton compared with the intact E. coli ribosome (Fig. 7). A simple explanation of this result is that FIG. 3. Sucrose gradient analysis for binding of rat P complex to the E. coli core ribosomes. The rat P complex was preincubated without (A) and with (B) 400 pmol of RNA fragments covering the GTPase-associated domain and then mixed with the E. coli core ribosomes, as described under "Materials and Methods." The ribosome samples were sedimented through a 10 -28% sucrose gradient. The graphs at the top show A 260 of portions of each fraction collected. Locations of 70 S ribosomes, and the 50 and 30 S subunits are indicated. Proteins in the fractions (a-f) were recovered by trichloroacetic acid-precipitation and analyzed by SDS-polyacrylamide gel electrophoresis, followed by immunoblotting using an anti-P monoclonal antibody (the panels at the bottom). L11, the protein responsible for efficient thiostrepton binding (35,38), is absent from a part of ribosome population. However, this is unlikely, because addition of excess L11 at the time of the hybrid formation gave no marked effect on the thiostrepton resistance (data not shown). It is more likely that the binding of the heterologous P complex affects the RNA conformation of the thiostrepton/L11 binding site. If this is the case, the slight conformational change of the RNA domain may take part in the conversion of the translocase accessibility from prokaryotic EF-G to eukaryotic EF-2. We infer that the stalk protein complex may play a role not only in interacting directly with the translocase, but also in modulating the functional conformation of rRNA.
Studies by three dimensional cryo-electron microscopy (39) and directed hydroxyl radical probing (40) suggest that EF-G interacts with several sites of the ribosome including the region at the base of the stalk. It is implied that major structural features of the translocase binding site are conserved between eukaryotic and prokaryotic ribosomes except for the stalk protein complex. In fact, two highly conserved regions, the GT-Pase-associated domain and the sarcin/ricin loop of E. coli 23 S rRNA have been identified as the EF-G binding sites by chemical footprinting analysis (4). The equivalent GTPase-associated domain (7) and the sarcin/ricin loop 2 of rat 28 S rRNA are also protected by binding of eukaryotic EF-2. A portion of the GTPase domain including residues 1056 -1103 of E. coli 23 S rRNA can be replaced with the homologous sequence of yeast 26 S rRNA without significant loss of the EF-G-dependent GTPase activity of the E. coli ribosome (41). Directed hydroxyl radical probing analysis by Wilson and Noller has clarified proximity relationships between EF-G and several other rRNA elements whose secondary structures are conserved between prokaryotes and eukaryotes (40). Therefore, we infer that the stalk protein complex and several conserved rRNA regions are the main determinants of the translocase-dependent function in the ribosome. The x-ray crystallographic studies have defined five domains of EF-G (42,43). Domains I and II are homologous to those of EF-Tu, and domains III, IV, and V mimic the size and shape of the tRNA portion of the EF-Tu-aminoacyl-tRNA-GTP ternary complex (44). The cryo-electron microscopic study on the ribosome-EF-G complex has shown that domains I and V are located at the base of the stalk and also that there is an "arc-like" connection between the stalk and the GЈ subdomain within domain I of EF-G (39). It seems to be likely that the kingdomdependent specificity of factor-ribosome interaction rests on the appropriate matching of the stalk protein complex with the GЈ domain of translocase.   4. Prokaryotic EF-G-and eukaryotic EF-2-dependent GTPase activity of the hybrid ribosome. Ribosome samples (2.5 pmol each), E. coli intact 70 S ribosomes (70S), E. coli ribosomes lacking L8 complex (70S-core), hybrid ribosomes constructed of E. coli core ribosomes and rat P complex (70S-P), and rat 80 S ribosomes (80S) were tested for their GTP hydrolysis activity depending on prokaryotic EF-G (A) and eukaryotic EF-2 (B), as described under "Materials and Methods".
FIG. 5. Induction of EF-2-dependent GTPase activity by mixing rat P complex and E. coli core ribosomes. Increasing amounts of rat P complex were preincubated with 2.5 pmol of E. coli core ribosomes (q) and without the core ribosome (Ⅺ) and their GTPase activity dependent on eukaryotic EF-2 was tested.
FIG. 6. RNA fragments of the GTPase domain prevent formation of the active hybrid ribosomes. Increasing amounts of RNA fragments encompassing the GTPase-associated domain of 23 S rRNA were preincubated with 0.5 g of rat P complex (q) and mixed with 2.5 pmol of E. coli core ribosomes and then assayed for the EF-2-dependent GTPase activity. Effect of addition of the RNA on the activity of the rat intact 80 S ribosomes was also tested (Ⅺ).

FIG. 7. Effect of thiostrepton on EF-2-dependent GTPase activity of the hybrid ribosomes.
Increasing amounts of thiostrepton were preincubated with 2.5 pmol of hybrid ribosomes (q), intact E. coli 70 S ribosomes (OE), and rat 80 S ribosomes (Ⅺ). The GTPase activity was assayed for individual samples.