Regulation of elongation factor G GTPase activity by the ribosomal state. The effects of initiation factors and differentially bound tRNA, aminoacyl-tRNA, and peptidyl-tRNA.

The elongation factor G (EF-G) is responsible for the translocation of the ribosome along the mRNA chain. Under in vitro conditions, EF-G exhibits a very active uncoupled GTPase activity which is dependent on the presence of ribosomes and is modulated by mRNA-dependent binding of tRNA. In the absence of tRNA, uncoupled EF-G GTPase is inhibited by initiation factors IF1 and IF3, but not by initiation factor IF2. In the presence of N-fMet-tRNAfMet and poly(A,U,G) or in the presence of N-acetyl-Phe-tRNAPhe and poly(U), initiation factor IF2 causes an additional decrease of the uncoupled EF-G GTPase activity. This effect, however, is dependent on the presence of IF1 and IF3 and is obviously due to the mRNA- and initiation factor-dependent binding of N-fMet-tRNAfMet and N-acetyl-Phe-tRNAPhe, respectively, to the ribosomal P-site. Non-enzymatic binding of N-fMet-tRNAfMet and N-acetyl-Phe-tRNAPhe, however, causes a stimulation of uncoupled EF-G GTPase activity. The same effects are observed for Met-tRNA, Phe-tRNAPhe and uncharged tRNA. These findings are discussed in the light of the three-site model of the ribosome and the mechanism of translocation.

Ribosome-dependent GTP hydrolysis by elongation factor (EF)'-G is required for the translocation of the ribosome along the mRNA chain (Nishizuka and Lipmann, 1966;Thach and Thach 1971;Gupta et al., 1971;Haselkorn and Rothman-Denes, 1973). Under i n uitro conditions, EF-G exhibits a very active uncoupled GTPase activity (Conway and Lipmann, 1964;Nishizuka and Lipmann, 1966;Kaziro et al., 1969). The uncoupling between EF-G GTPase and protein biosynthesis has greatly hampered the determination of a reliable stoichiometry between peptide bond formation and GTP hydrolysis (Nishizuka and Lipmann, 1966). Experimental evidence for the theoretically expected stoichiometry of 2 molecules GTP hydrolyzed per each new peptide bond has been obtained either by utilizing purified, endogenous polysomes and correction for the uncoupled EF-G GTPase activity (Cabrer et al., 1976) or by using rate limiting concentrations of EF-G (Chinali and Parmeggiani, 1980). ' The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The abbreviations used are: EF, elongation factor; IF, initiation factor. Several years ago, Kuriki and co-workers (Kuriki and Kanno, 1972;Kuriki and Yoshimura, 1974) showed that the ribosomal wash of Escherichia coli contains an inhibitor of uncoupled EF-G GTPase activity. These findings have induced us to study the coupling between EF-G GTPase activity and translocation of the ribosomes by investigating the mechanism by which this inhibitor reduces the uncoupled EF-G GTPase activity. Reinvestigating the results of Kuriki and co-workers (Kuriki and Kanno, 1972;Kuriki and Yoshimura, 1974), we have found two EF-G GTPase inhibitors in the ribosome wash of E. coli. One of these inhibitors has been purified to homogeneity and characterized (Voigt and Nagel, 1990). This factor inhibits both EF-G GTPase and poly(U)dependent poly(phenyla1anine) synthesis, but uncoupled EF-G GTPase activity is considerably more affected (Voigt and Nagel, 1990). Therefore, this inhibitor of uncoupled EF-G GTPase has been supposed to be a coupling factor (Voigt and Nagel, 1990). Furthermore, we have found that the 30 S ribosomal subunit is the target site of this inhibitor which consists of two polypeptides with apparent molecular masses of 10,000 and 23,000 (Voigt and Nagel, 1990). Recently, we have provided evidence that this inhibitor of uncoupled EF-G GTPase activity is a complex of initiation factors IF1 and IF3 (Nagel and Voigt, 1992). These findings have induced us to investigate whether or not the EF-G GTPase inhibitor activities are copurified with the initiation factors. Furthermore, we have systematically studied the effects of initiation factors and tRNA (peptidyl-tRNA, aminoacyl-tRNA and uncharged tRNA) on the uncoupled EF-G GTPase activity. The results are discussed on the basis of the recently developed allosteric three-site model for the ribosomal elongation cycle (Rheinberger and Nierhaus, 1986;Hausner et al., 1988;Gnirke et al., 1989).

EXPERIMENTAL PROCEDURES
Materials-ATP (Na+ salt), GTP (Lit salt) and a tRNA mixture from E. coli were purchased from Boehringer (Mannheim, Germany).
Preparation of Elongation Factors and Ribosornes-Electrophoretically homogeneous elongation factors G (EF-G) and T (EF-T) were prepared from E. coli B (Kaziro and Inoue, 1968;Sander et al., 1975). The EF-G and EF-T preparations used were free of ribosome-independent GTPase activities. Ribosomes, washed two or three times with 0.5 M NH&l, were separated into 30 S and 50 S subunits by sucrose gradient centrifugation at 0.5 mM Mg+ in a Beckman Ti-15 zonal rotor (Sander et al., 1975). Purity and activity of the subunits were tested routinely by analytical sucrose gradient centrifugation and poly(U)-dependent poly(pheny1ananine) synthesis. According to these tests, the 50 S subunits were maximally contaminated with 2-4% 30 S subunits, and the 30 S subunits contained 1-2% functional 50 S subunits. One AZe0 unit of ribosomal particles was taken to represent 25 pmol of 70 S, 39 pmol of 50 S, and 67 pmol of 30 S subunits (Sander et al., 1975). Protein concentrations were measured by the method of Lowry et al. (1951) using bovine serum albumin as standard.
Charging of tRNA-met-tRNA'"' was prepared by charging E.
Isolation of EF-G GTPase Inhibitors and Initiation Factors-The ribosome wash from several preparations of ribosomes (corresponding t o 1200 g of E. coli B) was combined (21 liters) and adjusted to 80% ammonium sulfate saturation by addition of solid (NH4)ZSO4. After 2 h at 4 "C, the precipitated proteins were collected by centrifugation at 20,000 x g for 20 min. Centrifugation and all of the following purification steps were performed a t 4 "C. The precipitated proteins were dissolved in a small volume of 20 mM Tris-HC1 (pH 7.5) containing 7 mM 2-mercaptoethanol, extensively dialyzed against the same buffer, and subjected to anion exchange chromatography using a Whatman DE-52 column (5.0 X 100 cm) equilibrated with 10 mM Tris-HC1 (pH 7.5). The column was washed with 300 ml of buffer A (20 mM Tris-HC1, pH 7.5, containing 7 mM 2-mercaptoethanol) and subsequently eluted with a linear gradient of 0-400 mM KC1 in buffer A (2 liters each). Fractions of 20 ml were collected and analyzed for absorbance a t 280 nm. Aliquots of 50 pl were taken from every third fraction and, after heating for 10 min at 55 "C, assayed for inhibitors of ribosome-dependent EF-G GTPase as recently described (Voigt and Nagel, 1990). Aliquots from the same fractions were analyzed (without preheating) for the initiation factors IFl, IF2, and IF3, respectively, by measuring the effect on the binding of N-[35S]fMet-tRNA'"' to ribosomes in the presence of poly(A,U,G) and the other two initiation factors as described by Parker Suttle et al. (1973). Two EF-G GTPase inhibitor peaks were eluted from the column as recently described (Voigt and Nagel, 1990). Both peak I and peak I1 contained IF1 and IF3 activities (Fig. 1). The inhibitor fractions of peak I were combined and directly applied to a phosphocellulose column (2.6 X 50 cm; Whatman P11, equilibrated with buffer B containing 20 mM Tris-HC1 (pH 7.5), 100 mM NH4C1, 1 mM EDTA, and 7 mM 2-mercaptoethanol). The inhibitor fractions of peak I1 were diluted with volume of buffer A before applied to a second phosphocellulose column (2.6 X 50 cm; also equilibrated with buffer B). T h e phosphocellulose columns were washed with 100 ml of buffer B and then eluted with linear gradients of 0.1-1.0 M NH4Cl. IF1 was eluted a t 0.5 M NH4C1 and IF3 at 0.65 M NH&1 (Fig. 2). Two peaks of EF-G GTPase inhibitor activities were obtained which were found to be coeluted with IF1 and IF3, respectively (Fig. 2). The active fractions were concentrated to 5-10 ml and finally purified by gel exclusion chromatography on columns (2.6 X 100 cm) of Sephadex G-75 and Sephadex G-100, respectively, equilibrated and eluted with buffer B. Again, EF-G GTPase inhibitor activities were found to be coeluted with initiation factors IF1 and IF3, respectively.
The initiation factor IF2 eluted from the DE-52 column ( Fig. 1) was further purified as described by Wahba and Miller (1974).
Binding or uncharged tRNA (250 pmol each) to ribosomes (80 pmol) was performed in the presence of 20 pg of poly (A,U,G) and 20 pg of poly(U), buffer 5 (5 mM MgClZ, 30 mM NH4C1, 30 mM KC1,2 mM dithiothreitol, respectively, in a total volume of 100 p1 containing ribosome standard 60 mM Tris-HC1, pH 7.8). After incubation for 12 min at 30 "C in the absence or presence of initiation factors or elongation factors (as indicated), the reaction mixtures were layered on 8-34% linear sucrose gradients in ribosome standard buffer 5 and centrifuged a t 4 "C for 2.5 h a t 38,000 rev/min in the Beckman SW40 T i rotor. The gradients were fractionated using an ISCO density gradient fractionator. The "70 s" peaks were collected and analyzed for bound tRNA and for their ability to induce EF-G GTPase activity.
GTPase Assays-EF-G GTPase activity was measured as the amount of 32Pi liberated during a 10-min incubation a t 30 "C (Sander et al., 1975). The reaction mixtures contained, unless otherwise stated, in a total volume of 75 pl: 4.5 pmol of Tris-HC1 (pH 7.8), 0.4 or 1.0 Fmol of MgCl,, 6 pmol of NH4C1, 0.2 pmol of dithiothreitol, 10 nmol of [ T -~~P I G T P (specific activity 150 cpm/pmol), 10 pmol of ribosomes, and 30 pmol of EF-G.

RESULTS
The Inhibitors of Uncoupled EF-G GTPase Present in the Ribosome Wash of E. coli-It has been shown that the ribosome wash of E. coli contains inhibitors of the uncoupled EF-G GTPase (Kuriki and Kanno, 1972;Kuriki and Yoshimura, 1974;Voigt and Nagel, 1990). One of these EF-G GTPase inhibitors stimulates poly(A,U,G)-and initiation factor 2 (IF2)-dependent binding of fMet-tRNAmet to ribosomes (Nagel and Voigt, 1992). Furthermore, partially purified initiation factors IF1 and IF3 have been shown to inhibit uncoupled EF-G GTPase activity (Nagel and Voigt, 1992). These findings have induced us to investigate whether the EF-G GTPase inhibitors are identical or merely copurified with the initiation factors IF1 and IF3.
As recently reported (Voigt and Nagel, 1990), DEAE-cellulose chromatography of the polypeptides present in the ribosome wash of E. coli have revealed two EF-G GTPase inhibitor peaks eluting behind the void volume and at about 70 mM KCl, respectively (Fig. 1,upper panel). Both IF1 and IF3 activities have been found to be copurified with both EF-G GTPase inhibitors (Fig. 1, lower panel). However, both peaks differ with respect to the relative IF1 and IF3 activity. Whereas the relative activities of EF-G GTPase inhibitor and initiation factors IF1 and IF3, respectively, have been found t o be constant in the different fractions of peak I, this is not always true for peak 11. A ribosome-independent GTPase activity has been found to be eluted from the DEAE-cellulose column at a salt concentration very similar to that for EF-G GTPase inhibitor peak I1 (data not shown). In some experiments, there was a partial overlap of the ribosome-independent GTPase activity with the EF-G GTPase inhibitor peak I1 simulating that initiation factors IF1/IF3 (peak 11) and EF-G GTPase inhibitor (peak 11) would be eluted at slightly different salt concentrations.
No EF-G GTPase inhibitor activity has been found in those fractions containing initiation factor IF2 (Fig. 1).
A complete separation of IF1 and IF3 activities has been observed when peak I1 eluted from the DEAE-cellulose column ( Fig. 1) has been subjected to phosphocellulose chromatography (Fig. 2, lower panel). Both the IF1 and IF3 peak contain EF-G GTPase inhibitor activities (Fig. 2, upper panel). Essentially the same results were obtained when peak I was subjected to phosphocellulose chromatography (data not shown). Subsequent gel exclusion chromatographies of initiation factors IF1 and IF3 on Sephadex G-100 and G-75 columns, respectively, again have revealed copurification of both initiation factors with EF-G GTPase inhibitor activities (data not shown), indicating that the EF-G GTPase inhibitors might be identical with initiation factors IF1 and IF3.
EF-G GTPase is differentially reduced by increasing concentrations of initiation factors IF1 and IF3, respectively (Fig.  3 ) . A considerably more pronounced inhibition of EF-G GTPase activity has been measured in the presence of both initiation factors than in the presence of IF1 or IF3 only (Fig.  3). However, a complete inhibition of EF-G GTPase activity has never been observed even at the highest concentrations of both initiation factors. The residual EF-G GTPase activity induced by 70 S ribosomes in the presence of saturating amounts of initiation factors/EF-G GTPase inhibitors is higher than EF-G GTPase activity which has been measured when 70 S ribosomes have been substituted by the same amounts of 50 S ribosomal subunits ( Table I). Inhibition of EF-G GTPase by IF1 and/or IF3 is more pronounced when 70 S ribosomes are preincubated in the presence of IF1 and/ or IF3 prior to the addition of EF-G than without preincubation (Table I). Furthermore, inhibition of 70 S ribosomeinduced EF-G GTPase by IF1 and/or IF3 can be overcome by addition of 30 S ribosomal subunits (Table I).
It has been shown that initiation factors IF1 and IF3 cause a dissociation of ribosomal subunits (Subramanian et al., 1968Sabol et al., 1970Subramanian and Davies, 1970;Dubnoff and Maitra, 1971;Miall and Tamaoki, 1972;No11 and Noll, 1972). Since the IF1-and IF3-induced dissociation of ribosomal subunits has been reported to be more pronounced at low than at high concentrations of divalent cations, we have investigated the inhibition of EF-G GTPase activity by IF1 and/or IF3 at different Mg2' concentrations. The inhibition of EF-G GTPase by IF3 but not the effect of IF1 can be reversed by increasing Mg2' concentrations ( Fig. 4).
Modulation of EF-G GTPase Activity by Binding of tRNA, Aminoacyl-tRNA, and Peptidyl-tRNA-The findings that initiation factors IF1 and IF3 inhibit uncoupled EF-G GTPase have induced us to investigate the modulation of EF-G GTPase activity by initiation factor-dependent and non-enzymatic binding of fMet-tRNANet and N-acetyl-Phe-tRNAPhe, respectively, to the ribosome. Since both the initiation factor-dependent and the non-enzymatic binding of fMet-tRNAfMet and N-acetyl-Phe-tRNAPhe, respectively, to the ribosome has been reported to be strongly dependent on

TABLE I
The inhibition of EF-G GTPuse activity by initiation factors IF1 and IF3: the effects of preincubation of ribosomes in the presence or absence of initiation factors Reaction mixtures for preincubation of ribosomes contained in a total volume of 50 pL 3.0 pmol of Tris-HCl (pH 7.8),4 pmol of NHICl, 0.3 pmol of MgC12, 0.2 pmol of dithiothreitol, 10 pmol of 70 S ribosomes, and (where indicated) 2 pg of IF1, 1 pg of IF3, 10 pmol of 50 S ribosomal subunits and 10 pmol of 30 S ribosomal subunits.
After 10 min at 30 "C (plus preincubation) or immediately after mixing (minus preincubation), the reaction mixtures were supplemented with standard buffer, 30 pmol of EF-G and further components as indicated and the reaction started by addition of 30 nmol of [ Y -~~P ] G T P (specific activity 150 cpm/pmol). The amounts of GTP hydrolyzed were determined as described by Sander et al. (1975) and eiven in nmol P; releasedL0 min. FIG. 4. Effect of MgCl, concentration on the ribosome-dependent EF-G GTPase activity in the absence or presence of initiation factors IF1 and/or IF3. 10 pmol of 70 S ribosomes were preincubated at 30 "C in a total volume of 50 pl of ribosome standard buffer in the absence or presence of initiation factors IF1 (6 pg) and/ or IF3 (4 pg) and subsequently analyzed for their ability to induce EF-G GTPase, as described under "Experimental Procedures." ., control; A, IF1; 0, IF3; V, IF1 plus IF3. the concentration of divalent cations, the effects on EF-G GTPase activity have been studied at different MgC12 concentrations (Fig. 5). All the experiments have been performed in the presence of poly(A,U,G) (Fig. 5, A and C ) and poly(U) (Fig. 5 , B and D ) , respectively.
As already reported in the literature (Chinali and Parmeggiani, 1982), addition of fMet-tRNAMet and N-acetyl-Phe-tRNAPhe, respectively, results in a stimulation of EF-G GTPase activity at low M$+ concentration and in an inhibition at high Mg2' concentrations in the absence of initiation factors. Stimulation of EF-G GTPase by binding of peptidyl-tRNA at low Mg2+ concentrations is almost reversed by addition of initiation factors IF1 and IF3 (Fig. 5 , A and B ) .
The inhibitory effects of initiation factors and peptidyl-tRNA on EF-G GTPase at high Mg2f concentrations, however, have been found to be additive (Fig. 5, A and B ) . An effect of initiation factor IF2 is observed only at low M e concentrations in the presence of IF1 and IF3 and either fMet-tRNAmet (in the presence of poly(A,U,G); Fig. 5 C ) or N-acetyl-Phe-tRNAPhe (in the presence of poly(U); Fig. 5 0 ) . Under these conditions, extremely low EF-G GTPase activities have been measured. These findings clearly demonstrate that EF-G GTPase activity is differentially affected by non-enzymatic and initiation factor-dependent binding of "peptidyl-tRNA," respectively, to the ribosomal P-site. At low MgClz concentrations, non-enzymatic binding of both aminoacyl-tRNA and peptidyl-tRNA takes place preferentially to the ribosomal Psite (Lucas-Lenard and Lipmann, 1967;Kaji et al., 1969). Under these conditions, EF-G GTPase activity is stimulated (Fig. 5, A and B ) . In the presence of all the three initiation factors, however, addition of either fMet-tRNANet plus poly (A,U,G) or N-acetyl-Phe-tRNAPhe plus poly(U) results in a very strong inhibition of EF-G GTPase (Fig. 5, C and D).
At 6 mM MgC12, an initiation factor-and poly(U)-dependent inhibition of EF-G GTPase has been observed for N-acetyl-Phe-tRNAPhe, but not for Phe-tRNAPh' or uncharged tRNAPhe (Table 11). In the absence of initiation factors, EF-G GTPase activity is stimulated at 6 mM MgClZ not only by addition of N-acetyl-Phe-tRNAphe, but also by addition of Phe-tRNAPhe or uncharged tRNA (Table   11). At 20 mM MgClz, EF-G GTPase activity is inhibited by the poly(U)-dependent binding of N-acetyl-Phe-tRNAPhe, Phe-tRNAPhe, or uncharged tRNA both in the presence and absence of initiation factors (Table 11).
Inhibition of EF-G GTPase by peptidyl-tRNA, aminoacyl-tRNA, and uncharged tRNA at high MgClZ is well known and has been attributed to the binding of tRNA to the ribosomal A-site (Modolell and Vazquez, 1973;Chinali and Parmeggiani, 1982). It is, therefore, not unexpected that this effect is almost unmodulated by the addition of initiation factors. Considerably more interesting are the differential effects of the mRNAdependent binding of peptidyl-tRNA to the ribosomal P-site observed at 6 mM MgC12 in the presence and absence of initiation factors ( Fig. 5; Table 11). Therefore, these effects

TABLE I1 T h e effects of synthetic mRNA and t R N A o n EF-G GTPase actiuity in the presence or absence of initiation factors
Reaction mixtures for preincubation of ribosomes contained in a total volume of 50 pl: 3.0 pmol of Tris-HCI (pH 7.8), 4.0 pmol of NH4C1, 0.3 or 1.0 pmol of MgC12, 0.2 pmol of dithiothreitol, 10 pmol of 70 S ribosomes, and, where indicated, 2 pg of IF1,l pg of IF2,1 pg of IF3,2.5 pg of poly(U), and 30 pmol of N-acetyl-Phe-tRNAPhe, Phe-tRNAPhe, or uncharged tRNAPhe. After a preincubation at 30 "C for 10 min, the reaction mixtures were supplemented with standard buffer, 10 pmol of EF-G and the reaction started by addition of 20 nmol of [Y-~*P]GTP (specific activity 150 cpm/pmol). The amounts of GTP hydrolyzed were determined as described by Sander et al. (1975) and given in pmol of Pi released/lO min. Regulation of EF-G GTPase Activity by the Ribosomal State have been studied at increasing EF-G concentrations (Fig. 6). Stimulation of EF-G GTPase activity by non-enzymatic, poly(U)-dependent binding of N-acetyl-Phe-tRNAPhe is more pronounced at low than at high EF-G concentrations (Fig.  6A). Double-reciprocal plots of these data (Fig. 6B) reveal that stimulation of EF-G GTPase activity by non-enzymatic binding of peptidyl-tRNA to the ribosomal P-site is not accompanied by an increase of Vmex, while the amount of EF-G needed for Vmarl2 is reduced. The inhibition of EF-G GTPase activity by initiation factors in the presence or absence of N-acetyl-Phe-tRNAPhe (Fig. 6A), however, results in a reduction of Vmax which is more pronounced in the presence than in the absence of N-acetyl-Phe-tRNAPhe (Fig. 6B). Furthermore, the amount of EF-G required for Vmar/P is reduced by the addition of initiation factors (Fig. 6B).
The Abilities of Different Ribosome/mRNA/tRNA Complexes to Induce EF-G GTPase-The ribosome-dependent EF-G GTPase activity measured in the presence of synthetic mRNA, peptidyl-tRNA, and/or initiation factors is induced by a mixture of modified and unmodified ribosomes. To eliminate the effects of unmodified 70 S ribosomes and free 50 S ribosomal subunits, we have isolated the different ribosome/mRNA/peptidyl-tRNA complexes (Fig. 7) and compared their abilities to induce EF-G GTPase activity at three different EF-G concentrations. All the different "physiological" ribosome/mRNA/peptidyl-tRNA complexes (Fig. 7) have revealed lower activities than unmodified 70 S ribosomes with  Table 111). The most striking differences between the isolated complexes and unmodified 70 S ribosomes have been observed at low EF-G concentrations (Table 111). Again, the most active ribosome species with respect to the induction of EF-G GTPase is the unphysiological complex I* formed by mRNA-dependent, non-enzymatic binding of peptidyl-tRNA, aminoacyl-tRNA, or uncharged tRNA to the ribosomal P-site at low Mg2+ concentration (Table 111), followed by the artificial complex IV* and the ''physiological'' complexes I11 and I V (Fig. 7). Complex I11 is the natural substrate for EF-G, the "pretranslocational" state of the ribosome formed by the action of the peptidyl transferase. As recently reported by Moazed and Noller (1989), this pretranslocational state is indeed a "hybrid state" of binding, because after the peptidyl transfer, the acceptor ends of the tRNAs have already moved spontaneously from the P-to the E-site and from the A-site to the P-site of the 50 S subunits, respectively. EF-G GTPase only promotes the second step of translocation, the movement of the anticodon ends of the tRNAs relative to the small ribosomal subunit, along with the mRNA (Moazed and Noller, 1989). The resulting complex I V (Fig. 7), is less active than complex I11 with respect to the induction of EF-G GTPase at low EF-G concentration ( Table   111). This difference, which is expected on the basis of the model of the ribosomal elongation cycle (Fig. 7), disappears, however, with increasing EF-G concentration. The artificial complex IV*, prepared by treatment of the isolated complex I V with puromycin, reveals a more pronounced induction of EF-G GTPase than complexes I11 and I V and unmodified 70 S ribosomes, especially when studied at low EF-G concentra- Abilities of different ribosome/mRNA/tRNA complexes to induce EF-G GTPase The different ribosome/mRNA/tRNA complexes were prepared by binding of N-acetyl-Phe-tRNAPh", Phe-tRNAPhe, and uncharged tRNAPh', respectively, to ribosomes in the presence of poly(U) and, where indicated, initiation factors and/or elongation factors. The reaction conditions, isolation of the complexes and GTPase assay conditions were as described under "Experimental Procedures." Complex I:  bound to ribosomes at 6 mM MgC12 in the presence of 2 pg of IF1, 1 pg of IF2, 1 pg of IF3, and 10 nmol of GTP; complex I*:  bound to ribosomes at 6 mM MgCI, in the absence of initiation factors; complex II*: prepared from complex I by a second incubation at 20 mM MgC1, (non-enzymatic binding of a second molecule of Nacetyl-Phe-tRNAPhe to the A-site); complex IP*: prepared from complex I by a second incubation at 20 mM MgCl, after addition of 300 pmol uncharged tRNAPhe; complex 111: prepared form complex I by second incubation after addition of 300 pmol of Phe-tRNAPhe and 30 pmol of EF-T; complex IV: prepared from complex I11 by a third incubation at 20 mM MgCI, after addition of 50 pmol of EF-G and 10 nmol of GTP; complex IV*: purified complex IV incubated in the presence of puromycin. GTP hydrolysis induced by 10 pmol of the different complexes or untreated 70 S ribosomes or 50 S ribosomal subunits was measured as described by Sander et al. (1975) (Table 111). The astonishing high activities measured in the presence of complex IV and especially in the presence of complex IV* have induced us to postulate a spontaneous transition of these complexes to the state of complex I11 (Fig.  7). The complexes I, II*, and II** (Fig. 7) have revealed the lowest EF-G GTPase activities, which are in the order of the activity of 50 S subunits (Table 111). Since after EF-T-dependent binding of aminoacyl-tRNA to the ribosomal A-site (complex 11) the peptidyl transfer occurs spontaneously (resulting in complex III), it is not possible to isolate complex 11. Therefore, the analogous complexes 11* and II** have been prepared by non-enzymatic binding of N-acetyl-Phe-tRNAPhe (complex II*) and uncharged tRNAPhe (complex II**) to complex 1 at 20 mM MgC12.

DISCUSSION
The inhibitors of uncoupled EF-G GTPase activity found in the ribosome wash of E. coli (Kuriki and Kanno, 1972;Kuriki and Yoshimura, 1974;Voigt and Nagel, 1990) are obviously identical with the initiation factors IF1 and IF3 as recently postulated (Nagel and Voigt, 1992). Our efforts to separate the EF-G GTPase inhibitor activities from these initiation factors have revealed that EF-G GTPase inhibitor activities are perfectly copurified with IF1 and IF3, respectively, under all chromatographic conditions tested (this pa- The initiation factors IF1 and IF3 have been reported to affect the association-dissociation equilibrium of the ribosomal subunits favouring the dissociation (Subramanian et al., 1968;Sabol et al., 1970;Subramanian and Davies, 1970;Dubnoff and Maitra, 1971;Miall and Tamaoki, 1972;Noll and Noll, 1972;Naaktgeboren et al., 1977). Dissociation of ribosomal subunits by initiation factors can be overcome by high M$+ concentrations (Subramanian et al., 1968). Inhibition of EF-G GTPase activity by IF3, but not the effect of IF1 can be reversed by increasing M$+ concentrations (Fig.  4). As shown by Naaktgeboren et al. (1977), IF1 acts by increasing the rate of exchange of the ribosomal subunits in the 70 S ribosome without changing the position of the equilibrium. Therefore, our finding that the effect of IF1 on EF-G GTPase activity cannot be reversed by high MgClz concentrations is in accordance with the literature.
The biological significance of the inhibition of EF-G GTPase by initiation factors IF1 and IF3 is obvious; in this way, uncoupled EF-G GTPase induced by 70 S ribosomes, which otherwise should take place also in vivo is presumably abolished (Table IV). A second mechanism to reduce uncoupled EF-G GTPase in vivo is, of course, the formation of ribosome/mRNA/tRNA complexes, i.e. the mRNA-and initiation factor-dependent binding of Met-tRNA""' to the ribosomal P-site (complex I; Fig. 7; Table 111) and the mRNA- and EF-Tu-dependent binding of aminoacyl-tRNA to the ribosomal A-site (complex 11; Fig. 7; Table 111). The relatively high activities observed for complex IV, which is the "product" of EF-G and was therefore expected to have a rather low ability to stimulate EF-G GTPase, is presumably not a problem in vivo. 1) At low EF-G concentrations, complex IV has a considerably lower ability to induce EF-G GTPase than complex 111, which is the "substrate" of EF-G ( Fig. 7; Table  111). 2) Under in vivo conditions, there is a competition between EF-G and EF-Tu for the binding to complex IV. Therefore, a stoichiometry of 2 molecules of GTP hydrolyzed per peptide bond formed (Cabrer et al., 1976;Chinali and Parmeggiani, 1980;Richter Dahlfors and Kurland, 1990;Voigt and Nagel, 1990) seems to be likely also for in vivo conditions. With respect to the mechanisms by which EF-G GTPase activity is regulated, the stimulation of EF-G GTPase activity by the non-enzymatic, but mRNA-dependent binding of peptidyl-tRNA, aminoacyl-tRNA, or uncharged tRNA to the ribosomal P-site (Chinali and Parmeggiani, 1982; this paper) is of special interest, although this effect is, of course, not physiological. Since this complex I* has a considerably higher affinity for EF-G than unmodified 70 S ribosomes or ribosome/mRNA complexes (Chinali and Parmeggiani, 1982;this paper), it seems to simulate the natural substrate of EF-G (complex 111; Fig. 7). Since the ability of the initiation complex (complex I; Fig. 7) to induce EF-G GTPase is rather low, whereas the activity of complex IV, which is the natural product, is astonishingly high (Fig. 7; Table 111), the E-site bound tRNA might have some effect on EF-G GTPase. One possible explanation for the relatively high ability of complex IV to induce EF-G GTPase activity might be a spontaneous back-transition to complex I11 (Fig. 7 ) , which in vivo is of course abolished by the EF-Tu-dependent binding of aminoacyl-tRNA to the ribosomal A-site. This is, however, no explanation for the high activity of complex I*. It seems to be reasonable to suppose that complex I* is a hybrid state (Moazed and Noller, 1989), which might be more stable than complex 111. Since a peptidyl transferase-catalyzed transfer of aminoacyl-and peptidyl-residues to puromycin takes place when aminoacyl-tRNA and peptidyl-tRNA, respectively, are non-enzymatically bound to the ribosomal P-site, we assume that complex I* contains 1 tRNA molecule which is P-site bound to the 50 S subunit and A'-site bound to the 30 S subunit. This complex I* strongly stimulates EF-G GTPase activity but is obviously not converted to a complex I state by the action of EF-G. This model implies that the E-site/P'site bound uncharged tRNA is required for the EF-G-dependent translocation of the ribosome along the mRNA chain.