Functional Interactions between the G′ Subdomain of Bacterial Translation Factor EF-G and Ribosomal Protein L7/L12*

Protein L7/L12 of the bacterial ribosome plays an important role in activating the GTP hydrolytic activity of elongation factor G (EF-G), which promotes ribosomal translocation during protein synthesis. Previously, we cross-linked L7/L12 from two residues (209 and 231) flanking α-helix AG′ in the G′ subdomain of Escherichia coli EF-G. Here we report kinetic studies on the functional effects of mutating three neighboring glutamic acid residues (224, 228, and 231) to lysine, either singly or in combination. Two single mutations (E224K and E228K), both within helix AG′, caused large defects in GTP hydrolysis and smaller defects in ribosomal translocation. Removal of L7/L12 from the ribosome strongly reduced the activities of wild type EF-G but had no effect on the activities of the E224K and E228K mutants. Together, these results provide evidence for functionally important interactions between helix AG′ of EF-G and L7/L12 of the ribosome.

Protein L7/L12 of the bacterial ribosome plays an important role in activating the GTP hydrolytic activity of elongation factor G (EF-G), which promotes ribosomal translocation during protein synthesis. Previously, we cross-linked L7/L12 from two residues (209 and 231) flanking ␣-helix A G in the G subdomain of Escherichia coli EF-G. Here we report kinetic studies on the functional effects of mutating three neighboring glutamic acid residues (224, 228, and 231) to lysine, either singly or in combination. Two single mutations (E224K and E228K), both within helix A G , caused large defects in GTP hydrolysis and smaller defects in ribosomal translocation. Removal of L7/L12 from the ribosome strongly reduced the activities of wild type EF-G but had no effect on the activities of the E224K and E228K mutants. Together, these results provide evidence for functionally important interactions between helix A G of EF-G and L7/L12 of the ribosome.
Elongation factors (EF) 3 Tu and G are the bacterial counterparts of universal translation factors, members of the G protein superfamily (1), which hydrolyze GTP to GDP and inorganic phosphate (P i ). The ribosome activates the latent GTPase activities of EF-Tu and EF-G. These factors, in turn, regulate protein synthesis by promoting specific molecular movements in the ribosome during protein synthesis. EF-Tu delivers aminoacyl-tRNA substrates to the ribosome, dependent on codon-anticodon pairing. EF-G promotes ribosomal translocation, involving movement of two tRNAs and mRNA in the ribosomal cavity, following the formation of each peptide bond.
GTP hydrolysis by EF-Tu and EF-G occurs on their G domains, which are similar to one another in their amino acid sequences and their core tertiary structures. The hydrolysis reaction involves in-line nucleophilic attack of a water molecule on the ␥-phosphorus of GTP (2). On EF-Tu (and possibly also EF-G), this reaction is catalyzed by a conserved histidine residue (3), whose side chain is believed to rotate to a position next to the water molecule (4). In contrast to EF-Tu, EF-G contains a GЈ subdomain, which is invariably inserted between ␣-helices D G and E G (5). The GЈ subdomain is also present in the same location in EF-2, the eukaryotic cytoplasmic homolog of EF-G (6). However, the GЈ subdomains of EF-G and EF-2 are unrelated in their amino acid sequences and tertiary structures. The functional significance of the GЈ subdomain in either factor has not been determined.
How the bacterial ribosome activates GTP hydrolysis by EF-Tu and EF-G remains obscure. Early research identified L7/L12, 4 a protein component of one of the peripheral stalks of the ribosome, as an important contributor to GTPase activation of both factors (7,8). More recent studies identified residues of the C-terminal domain (CTD) of L7/L12, important for GTPase activation and rapid association of both factors with the ribosome (9, 10). Nucleotides of 23 S RNA (11) and unidentified ribosomal components (12) have also been implicated in GTPase activation of these factors.
The sites on EF-Tu and EF-G that interact with L7/L12 are only beginning to be elucidated. For EF-Tu, mutational studies identified negatively charged residues in its helix D G that may interact with positively charged residues of L7/L12 (9). For EF-G, no functional studies of this type have been reported so far. However, two topographic studies indicated that the GЈ subdomain of EF-G is proximal to L7/L12. First, residues 209 and 231, flanking helix A GЈ , of Escherichia coli EF-G were crosslinked to L7/L12 (13). In the same study, no cross-links were detected from two other EF-G residues (156 and 160) of helix D G , whose sequence is completely different from the corresponding helix of EF-Tu. Second, a cryo-EM study localized EF-G residue 209 near the base of the ribosomal stalk, which was interpreted as an interaction with the CTD of L7/L12 (14).
This study was undertaken to investigate the potential functional roles of the GЈ subdomain of EF-G. We present results from kinetic experiments that provide evidence for interactions between residues of helix A GЈ of EF-G and L7/L12, which are important for activating GTP hydrolysis and ribosomal translocation.

EXPERIMENTAL PROCEDURES
Materials-E. coli 70 S ribosome and phage T4 gene 32 mRNA were prepared as described (15). L7/L12 was specifically stripped from the ribosome by an ethanol-NH 4 Cl washing procedure (12). The extent of L7/L12 stripping was assessed by immunoblotting using a L7/L12 polyclonal antibody (13). E. coli tRNA fMet and tRNA Phe were purchased from Sigma. The mRNA, 5Ј-AAGGAGGUAAAAAUGUUUGCU(N 6 )-3Ј, was synthesized by Dharmacon and conjugated at its 3Ј end with pyrene as described (16). Mant-labeled nucleotides were purchased from Invitrogen.
GTP Binding and Hydrolysis Assays-All fluorescence measurements described below were made by a QM-6 fluorimeter (Photon Technology International). Binding of mant-GTP to EF-G was measured by titrating mant-GTP (1 M) in 2 ml of buffer 2 with EF-G (0 -15 M). After adding each EF-G aliquot, mixing with a stir bar, and allowing the solution to reach equilibrium, fluorescence (excitation 362 Ϯ 2 nm; emission 438 Ϯ 4 nm) was recorded.
Multiple-turnover GTP hydrolysis was assayed as described (13), with the following modifications. EF-G (0.04 M) and ribosomes (variable concentration) were preincubated (37°C, 10 min) in 9 l of buffer 2. Reactions (10 l) were started by adding 1 l of 500 M GTP (containing 0.05 Ci of [␥-32 P]GTP). Samples (2 l) were withdrawn after appropriate time intervals, during the linear kinetics of the reaction. Samples were quenched and analyzed by TLC as described (13). Data of GTP molecules hydrolyzed per EF-G molecule/s (v o ) as a function of ribosome concentration [R] were calculated and fitted (via Sig-maPlot2000 software) to: ). Multiple-turnover mant-GTP hydrolysis was detected by the fluorescence change between mant-GTP and mant-GDP bound to EF-G. The fluorescence of mant-GTP (120 M) in 250 l of buffer 2 was monitored by exciting the fluorophore at 362 nm and detecting its emission at 438 nm. EF-G (25 M) and vacant ribosomes (0.7 M) were added in successive steps and mixed by pipetting. Samples (20 l) were removed from the cuvette and precipitated with 0.2 M HCl. The supernatants of the samples were concentrated in a vacuum centrifuge to ϳ8 l, and 2 l of each sample (ϳ500 pmol) was analyzed by TLC (17).
Single-turnover mant-GTP hydrolysis was monitored in a stopped-flow device (MiniMixer, KinTek), which contained two reactant syringes. Syringe A contained 1 ml of 24 M vacant ribosomes in buffer 2 at ϳ22°C. Syringe B contained 1 ml of 20 M EF-G and 10 M mant-GTP (same buffer and temperature).
In some experiments, AlF 4 Ϫ complex (100 M) was included in syringe B. Samples (ϳ200 l) from each syringe were rapidly mixed together (dead time of ϳ3.5 ms) and injected into a cuvette in our fluorimeter. Fluorescence (excitation 362 Ϯ 2 nm; emission 438 nm Ϯ 4) was monitored over 40 -400 s, with measurements taken every 50 -1000 ms, depending on the rate constant of the EF-G protein being tested. Data for each reaction were fitted to a single exponential equation, F t ϭ F ∞ ϩ ⌬F max ϫ exp(Ϫk obs ϫ t), where F 0 is the initial fluorescence at time ϭ 0; F t is the fluorescence at time t; ⌬F max is the maximum fluorescence change (F 0 Ϫ F ∞ ), and k obs is the observed reaction rate constant.
Ribosomal Translocation Assays-Multiple-turnover ribosomal translocation was monitored by the toeprinting method (18). A pretranslocation complex (1 M) was formed in 90 l of buffer 3 with E. coli ribosomes (1 M) containing phage T4 gene 32 mRNA (0.8 M) and uncharged tRNA f Met (1.2 M) and tRNA Phe (1.2 M) bound in the P and A sites, respectively, of the ribosome (15). EF-G (0.1 M) and GTP (500 M) were added, and reactions were incubated at 37°C. Samples (10 l) were removed after various time intervals (as indicated) and analyzed by toeprinting (15).
Single-turnover ribosomal translocation was monitored by the change in fluorescence of mRNA labeled with pyrene at its 3Ј end (16). A pretranslocation complex (0.5 M) was formed in 5 ml of buffer 3, similar to the above complex except substituting pyrene-mRNA (0.4 M). This complex was divided into five equal aliquots, which were loaded successively into syringe A of the stopped-flow device. Syringe B contained the various EF-G proteins (1.25-12 M) and GTP (2 mM) in 1.2 ml of buffer 3. Reactions were started by rapidly mixing 200 l samples from each syringe. Fluorescence (excitation 332 Ϯ 4 nm; emission 377 Ϯ 8 nm) was monitored over 200 -300 s, with averaged measurements taken every 60 -200 ms, depending on the rate constant of the EF-G protein being tested. Data were fitted to a double exponential equation: F t ϭ F ∞ ϩ ⌬F 1 ϫ exp(Ϫk 1 ϫ t) ϩ ⌬F 2 ϫ exp(Ϫk 2 ϫ t). Observed translocation rates as a function of EF-G concentration were fitted to Reaction 1.

RESULTS
To investigate the function of the GЈ subdomain, we first genetically replaced the entire subdomain of E. coli EF-G (residues 166 -261) with a single Gly residue. The N-and C-terminal ends of the GЈ subdomain are closely juxtaposed in the x-ray crystal structures of Thermus thermophilus EF-G proteins (5,19), suggesting that this deletion would not perturb the folding of the rest of EF-G. We introduced this genetic deletion into a plasmid encoding E. coli EF-G with a C-terminal His 6 tag (15). Although the mutant protein could be produced at high levels in E. coli, it aggregated in cellular inclusion bodies, and we were unable to refold it after purification under denaturing conditions. We then replaced single amino acid residues of the GЈ subdomain. In selecting these mutations, we considered three sources of information as follows: (i) our previous cross-linking results indicating the proximity of L7/L12 to EF-G residues 209 and 231, which flank helix A GЈ (13); (ii) an alignment of the amino acid sequences of the GЈ subdomain from bacterial and mitochondrial phyla (supplemental Fig. 1); and (iii) a study identifying conserved, positively charged residues of L7/L12 important for EF-G GTPase activity (10), suggesting that L7/L12 interacts with conserved, negatively charged residues of EF-G.
The GЈ subdomain is generally poorly conserved in relation to other EF-G domains. We chose to mutate the three most conserved, negatively charged residues corresponding to Glu-224, Glu-228, and Glu-231 of E. coli EF-G. In the structures of T. thermophilus EF-G proteins, residues equivalent to Glu-224 and Glu-228 are located within helix A GЈ , and the residue equivalent to Glu-231 is at the N terminus of helix B GЈ (Fig. 1). The side chains of all three residues are exposed to solvent. We replaced these Glu residues with oppositely charged Lys residues, either individually or in combination, hoping that these mutations would repel interactions with L7/L12. The His 6 -tagged EF-G proteins (four GЈ mutants and wild type) were produced and purified under native conditions (15).
GЈ Mutations Hinder Ribosome-activated GTP Hydrolysis-We first examined the GTP hydrolysis activities of the purified EF-G proteins, activated by vacant ribosomes (lacking tRNA and mRNA). Experiments were performed under multipleturnover conditions of fixed EF-G concentration (0.04 M), saturating GTP concentration (500 M), and increasing ribosome concentration. As shown in Fig. 2A, the initial velocity (v o ) of GTP hydrolysis catalyzed by wild type EF-G increased with increasing ribosome concentrations up to ϳ2 M, followed by a drop in v o at higher ribosome concentrations, because of turnover inhibition effects as reported previously (20). Based on fitting the Michaelis-Menten equation to the data up to 2 M ribosome, wild type EF-G and the E231K mutant displayed nearly the same turnover (k cat ϭ 9.9 and 7.7 s Ϫ1 , respectively) and apparent ribosome binding (K m ϭ 0.91 and 1.5 M). In contrast, the other three GЈ mutants, E224K, E228K, and the triple Lys mutant (TK), were severely defective, to the extent that their v o values were not measurable at any ribosome concentration above a control reaction containing only the ribosome.
The latter three GЈ mutants were specifically defective in ribosome-activated GTP hydrolysis. In the absence of the ribosome, the basal GTPase activities of all four GЈ mutants varied between ϳ3 ϫ 10 Ϫ4 and 3 ϫ 10 Ϫ3 s Ϫ1 , slightly faster than ϳ4 ϫ 10 Ϫ4 s Ϫ1 for the wild type protein, at the same EF-G and GTP concentrations indicated above (data not shown). Under multiple-turnover conditions, the defects in ribosome-activated GTP hydrolysis could, in principle, be manifested at one or several kinetic steps in the uncoupled GTPase reaction cycle (20). We could exclude effects on the initial binding of GTP to EF-G, as all four GЈ mutants bound a fluorescent substrate derivative, mant-GTP, with affinities (K d values: 2.2-3.7 M) that were indistinguishable from the wild type (Fig. 2B).

FIGURE 1. Locations of mutated residues in the G subdomain of EF-G.
Schematic is based the structure of T. thermophilus EF-G-2 bound to GTP; Protein Data Bank accession code 1WDT (19). Yellow, GЈ subdomain; orange, residues equivalent to E. coli Glu-224, Glu-228, and Glu-231. To try to localize the defects of the GЈ mutants to specific steps in the reaction cycle, we sought a GTPase kinetic assay performed under single-turnover conditions. Using mant-GTP as a fluorescent reporter, we observed that its fluorescence increased substantially when wild type EF-G was added (Fig.  3A), as reported previously (21). When we subsequently added a limiting amount of the ribosome, the fluorescence gradually dropped to a lower plateau. This suggested that mant-GTP was slowly hydrolyzed to mant-GDP, although it was bound to EF-G, activated by a limiting amount of ribosome that was recycled.
To test this hypothesis, we removed three samples (S1, S2, and S3) at different time points during the fluorescence decay. These samples were immediately quenched by acid precipitation and subsequently analyzed by TLC (Fig. 3B). According to  the mobilities of known molecular markers resolved on the same TLC, S1 (taken after EF-G addition) contained only mant-GTP; S2 (taken after ribosome addition, near the end of the fluorescence decay) contained mostly mant-GDP and small amounts of mant-GTP, and S3 (taken in the fluorescence plateau) contained only mant-GDP. Thus, these experiments clearly show that the fluorescence decrease was because of mant-GTP hydrolysis.
The fluorescence decrease of the mant group presumably reflects a conformational change in EF-G. It is unclear whether this occurs during cleavage of the phosphoanhydride bond or during the subsequent release of the P i product from EF-G. To characterize the mechanism further, we monitored singleturnover mant-GTP hydrolysis by using a stopped flow device that rapidly mixed EF-G⅐mant-GTP with a stoichiometric amount of the ribosome. We compared reactions in the presence or absence of AlF 4 Ϫ , an analog of the planar transition state structure of the ␥-phosphate leaving group, which binds tightly to many G protein⅐GDP complexes.
Under single-turnover conditions, the fluorescence signal decayed exponentially with much faster kinetics and similar in amplitude (Fig. 3C). When AlF 4 Ϫ was included with EF-G⅐mant-GTP before mixing them with the ribosome, the rapid fluorescence decrease was followed by an increase, which was slower and smaller in amplitude. As a positive control, when EF-G⅐mant-GDP and AlF 4 Ϫ were mixed with the ribosome, only a larger fluorescence increase was observed. As a negative control, when EF-G⅐mant-GDP was mixed with the ribosome, a smaller fluorescence increase was observed. Collectively, these results suggest that the initial fluorescence decrease encompasses both mant-GTP hydrolysis and P i release from EF-G. The subsequent fluorescence increase is because of AlF 4 Ϫ binding to EF-G⅐mant-GDP, which follows P i release. The small fluorescence increase seen in the negative control most likely reflects binding of EF-G⅐mant-GDP to the ribosome.
We then applied the above assay to compare the GЈ mutants and wild type EF-G proteins in catalyzing mant-GTP hydrolysis (Fig. 4). Under single-turnover conditions, wild type EF-G cat-alyzed this reaction with k obs of 1.8 s Ϫ1 (Table 1). This value is ϳ10-fold slower than previously reported for P i release (20 s Ϫ1 ), and much slower than GTP hydrolysis (80 s Ϫ1 ), as measured by other methods under comparable conditions (22). These comparisons further suggest that the fluorescence decrease is not caused by the hydrolysis reaction itself, but rather because of a slower EF-G conformational change that follows P i release. Alternatively, the slower rate may arise from the mant group partially interfering with the hydrolysis reaction.
Regardless of the precise explanation, the GЈ mutants showed substantially reduced rates of mant-GTP hydrolysis, in comparison to wild type EF-G ( Table 1). The most defective single GЈ mutants were E224K and E228K, whose k obs values were both 65-fold lower than wild type. The E231K mutant was the most active, with k obs only 2-fold lower than wild type, in accord with the multiple-turnover GTP hydrolysis assay ( Fig.  2A). The TK mutant was the most defective of all, with k obs that was 200-fold lower than wild type. However, the TK mutant still displayed significant ribosome-activated GTPase activity, relative to controls of the ribosome alone (Fig. 4) and the basal GTPase activity of the TK mutant in the absence of the ribosome.
In summary, results from multiple-and single-turnover assays both identified mutants E224K and E228K as having large defects in GTP hydrolysis. The single-turnover assays provide the more direct and accurate measurements, but they are ambiguous with regard to effects on hydrolysis or P i release. It is also possible that the apparent defects in mant-GTP hydrolysis are manifested because of an impaired association of GЈ mutants with the ribosome. The latter possibility is addressed below.
GЈ Mutations Reduce Ribosomal Translocation Kinetics-Previous studies have indicated that GTP hydrolysis by EF-G is strongly coupled to ribosomal translocation, the coordinated movement of tRNA and mRNA in the ribosomal cavity. Upon binding of EF-G⅐GTP to a pretranslocational ribosome complex, GTP hydrolysis occurs rapidly (23), which is thought to induce conformational rearrangements in the complex that drive the subsequent steps of P i release from EF-G⅐GDP and translocation (22,24). Although translocation can occur in the presence of nonhydrolyzable GTP analogs, GTP hydrolysis strongly promotes translocation both kinetically and thermodynamically (24,25).
Given the effects of the GЈ mutations on GTP hydrolysis activated by the vacant ribosome, we reasoned that these mutations might also exert effects on translocation if, as in the wild type situation, GTP hydrolysis and translocation remain strongly coupled mechanisms. We assembled a pretransloca-   tional ribosome complex that contained a model mRNA and two tRNAs bound in the A and P sites of the ribosome (15). We first measured translocation in the toeprinting assay, which monitors the position and movement of the ribosome relative to its bound mRNA, as detected by reverse transcription of a DNA primer annealed to the mRNA downstream of the ribosome (18). Because reverse transcription is much slower than ribosomal translocation, we examined translocation reactions under multiple-turnover conditions of limiting EF-G (0.1 M), saturating GTP (500 M), and pretranslocational complex (1 M). Fig. 5 compares ribosomal translocation profiles catalyzed by wild type EF-G and the TK mutant. At the start of reactions, the pretranslocational complex was indicated by two primer extension products, a major one stopping at ϩ17 and minor one at ϩ16, relative to the initiation codon of the mRNA. Following addition of wild type EF-G and GTP, the ϩ17 product was converted rapidly (within the 1st min) to a ϩ19 product, indicating the post-translocational complex. The ϩ16 product was converted more slowly to the same ϩ19 product. Following addition of TK and GTP, both ϩ16 and ϩ17 products were converted to the ϩ19 product at about the same kinetics and much slower than the wild type reaction. The E224K and E228K mutants catalyzed slower translocation similar to the TK mutant, whereas E231K catalyzed faster translocation between the rates of wild type and TK (supplemental Fig. 2).

؉EF-G
To measure accurately the kinetics of translocation under single-turnover conditions, we took advantage of a fluorescence assay that monitors the movement of pyrene attached to the 3Ј end of a short mRNA that just spans the ribosome (16). A pretranslocational complex (0.5 M) containing pyrene-labeled mRNA and the same two tRNAs was rapidly mixed with wild type EF-G (1.25 M) and saturating GTP. The fluorescence decay associated with translocation followed distinctly biphasic kinetics (supplemental Fig. 3). In the first phase of the reaction, k obs was 0.58 s Ϫ1 , whereas k obs of the second phase was 0.056 s Ϫ1 . It should be noted that the rate constant of the faster phase is comparable with the original report (16) but slower than reported in other studies (22,26). This discrepancy has been recently attributed to the C-terminal His 6 tag, which is present in the EF-G proteins of our study, and other smaller factors (26).
Nonetheless, the GЈ mutants promoted single-turnover translocation at rates that were significantly slower than wild type ( Fig. 6; Table 1). Their relative rates were in the same order as measured in single-turnover mant-GTP hydrolysis. However, the quantitative differences between wild type and GЈ mutants were smaller in translocation (e.g. ϳ10-fold between wild type, E224K, and E228K). The biphasic kinetic behavior was retained in all reactions. The GЈ mutants were largely affected in the first reaction phase, and only slight affected in the second phase.
To dissect these effects further, we monitored translocation rate as a function of EF-G concentration, which was always in   excess over pretranslocational complexes (i.e. pseudo-unimolecular conditions). The k obs of the first phase of translocation was strongly dependent on the concentration of wild type EF-G up to ϳ6 M (Fig. 7), whereas the slower phase was only slightly dependent (data not shown).
These results indicated that the first phase of translocation was rate-limited by the bimolecular association of EF-G with the ribosome. Compared with the wild type behavior, translocation catalyzed by all four GЈ mutants was almost unaffected by their concentrations. Thus, the differences between wild type and the GЈ mutants became even greater at the highest EF-G concentrations we tested. These results argue against activity defects of the GЈ mutants being due to their impaired, productive association with the ribosome. Rather, they suggest that the GЈ mutants are ratelimited by the subsequent (unimolecular) step of GTP hydrolysis, which leads to reduced ribosomal translocation kinetics.
GЈ Mutant Defects Result from Disrupted Interactions with L7/L12-To examine whether the defects of the GЈ mutants result from their disrupted interactions with L7/L12, we specifically stripped L7/L12 from the ribosome by an ethanol-NH 4 Cl washing procedure (12). This procedure removed ϳ95% of L7/L12 while leaving other proteins on the ribosome intact (Fig.  8). We then compared the GTPase and ribosomal translocation activities of the GЈ mutants and wild type EF-G.
Removal of L7/L12 from the ribosome reduced the rate of GTP hydrolysis by wild type EF-G by a factor of ϳ140 (from 9.9 to 0.070 s Ϫ1 ) under multiple-turnover conditions (data not shown). In mant-GTP hydrolysis by wild type EF-G under single-turnover conditions, the effect of removing L7/L12 from the ribosome was smaller, ϳ12-fold (Table 1), which again suggests that the mant group partially interferes with GTP hydrolysis. In contrast, the most defective GЈ mutants (E224K, E228K, and TK) catalyzed mant-GTP hydrolysis at nearly the same rates either in the presence or the absence of L7/L12. In the absence of L7/L12, these mutants and wild type EF-G became closer in their relative activities, although the wild type protein remained the most active. E231K displayed a behavior between the most severe GЈ mutants and wild type EF-G.
L7/L12 effects on ribosomal translocation mirrored those measured in GTP hydrolysis assays. Removal of L7/L12 from the ribosome had no effect on the abilities of the most defective GЈ mutants to promote translocation. The largest effect of L7/L12 removal was again observed in the wild type EF-G reactions. Consequently, in the absence of L7/L12, all four EF-G proteins catalyzed translocation with similar kinetics (Table 1). In summary, the two sets of data in Table 1 provide strong evidence that the defects of the GЈ mutants result largely from disrupted interactions with L7/L12.

DISCUSSION
L7/L12 is one of a few bacterial ribosomal proteins of which accumulated evidence points to its functional roles in protein synthesis, whereas other ribosomal proteins appear to play structural roles in folding and modulating the conformations of the ribosomal RNAs, which catalyze peptide bond formation and decoding. Studies during the 1970s first demonstrated the requirement of L7/L12 for the functions of a wide variety of essential translation factors, including G and non-G proteins, in protein synthesis in vitro (7,8,27,28). Subsequent studies isolated mutants of L7/L12, which altered the fidelity and speed of protein synthesis in vivo (29). Recent in vitro studies have converged on a small cluster of conserved, positively charged or aliphatic residues of L7/L12, which may interact with the G protein factors EF-Tu, EF-G, RF3, and IF2 (9,10,30).
From the latter studies, one might have anticipated that L7/L12 would interact with the G domains that are shared among the G protein factors. Instead, other studies (summarized in the Introduction) suggested that L7/L12 interacts with helix D G of EF-Tu and helix A GЈ of EF-G (9,13,14). These two helices have been proposed to contact a common (unspecified) component of the ribosome (31). In RF3 a portion of the GЈ subdomain, including helix A GЈ , is present (32), but in IF2 the GЈ subdomain is completely missing, which suggests that these factors also may interact distinctly with L7/L12. This idea has precedent in other G proteins, which interact with GTPaseactivating proteins (GAPs) in structurally distinct ways (33).
Functional Interactions between the GЈ Subdomain and L7/L12-Our present study provides evidence for functionally important interactions between helix A GЈ of EF-G and L7/L12. Of the three single mutants we characterized, E224K and E228K exhibited the largest defects in EF-G activities on the ribosome. These mutations involve conserved Glu residues located near the C-terminal end of helix A GЈ . These GЈ mutants were most severely defective in ribosome-activated GTP hydrolysis, whereas their basal GTPase activities in the absence of the ribosome were not affected.
Consistent with the notion that these mutations disrupt electrostatic interactions with L7/L12, we found that removing L7/L12 from the ribosome had little or no effect on the GTPase activities of the E224K and E228K mutants. In contrast, removal of L7/L12 from the ribosome resulted in a substantial reduction in the GTPase activity of wild type EF-G, in accord with previous studies (7,8,10,12). From these results, we anticipate that residues Glu-224 and Glu-228 in helix A GЈ are involved in ion pairs with positively charged residues of L7/L12, which have been identified previously (10). By analogy to EF-Tu (9), several conserved hydrophobic residues adjacent to Glu-224 and Glu-228 in helix A GЈ may also interact with L7/L12 and thereby stabilize the electrostatic interactions.
The same order of activity of EF-G proteins (TK Ͻ E224K Ϸ E228K Ͻ E231K Ͻ wild type) was found in GTP hydrolysis and ribosomal translocation assays. The quantitative effects were uniformly larger in GTP hydrolysis, relative to translocation. This suggests that the GЈ mutations exert direct effects on GTP hydrolysis, which in turn leads to indirect effects on translocation (23)(24)(25). Alternatively, the differences in the observed effects on GTP hydrolysis and translocation may simply reflect a differential interference by the fluorescent probes or the His 6 tag on EF-G (see "Results"). Our mant-GTP hydrolysis experiments, which monitor a conformational change in EF-G, suggest that the GЈ mutations perturb GTP hydrolysis and/or P i release from EF-G. These results are in partial accord with a study analyzing mutations in the CTD of L7/L12, which reported effects on P i release but surprisingly no significant effects on translocation (34).
Implications-Several questions remain unanswered. First, L7/L12 is the only protein in bacterial ribosomes that is present in multiple copies, which varies depending on species (10). Thus, it is unclear whether EF-G interacts with single or multiple L7/L12 proteins.
Second, it is unclear how interactions of L7/L12 with the GЈ subdomain leads to activation of GTP hydrolysis on the G domain, some 35 Å away from helix A GЈ . Possibly, these interactions may allosterically alter the configurations of catalytic residues of the G domain, or may affect the positioning of another ribosomal component that directly contacts the GTP substrate. According to cryo-EM models, nucleotides of helix 95 (or Sarcin-Ricin loop) of 23 S RNA may directly contact the G domain (19,35).
Finally, in translation systems in vitro, EF-G and EF-Tu are not interchangeable with their eukaryotic cytosolic homologs, EF-2 and EF-1␣, respectively (36). This is despite the high degree of sequence conservation in these factors and ribosomes. One major difference is that the eukaryotic factors interact with P proteins, unrelated to L7/L12, on the stalk of their cognate ribosomes (37). Experiments in which L7/L12 was replaced with P proteins on the E. coli ribosome demonstrated functionality of the eukaryotic factors on these hybrid ribosomes (38). Considered in light of our present results, we speculate that the specificity determinants of EF-G and EF-2 may involve the GЈ subdomains of these factors, which have unrelated sequences and tertiary structures (6). The specificity determinants of EF-Tu and EF-1␣ remain unknown, but it is interesting to note that their G domains differ by the addition or deletion of several ␣-helices (39), and their shared D G helices have different sequences.