Direct Evidence of an Elongation Factor-Tu/Ts·GTP·Aminoacyl-tRNA Quaternary Complex*

Background: Elongation factor-Tu (EF-Tu) chaperones aminoacyl-tRNA (aa-tRNA) to the elongating ribosome as a ternary complex with GTP. Results: EF-Ts facilitates EF-Tu·GTP binding to aa-tRNA through direct, transient interactions. Conclusion: EF-Ts accelerates the formation and decay of ternary complex through the formation of a transient EF-Tu/Ts·GTP·aa-tRNA quaternary complex. Significance: These newly described interactions of EF-Ts may serve to regulate ternary complex abundance in the cell. During protein synthesis, elongation factor-Tu (EF-Tu) bound to GTP chaperones the entry of aminoacyl-tRNA (aa-tRNA) into actively translating ribosomes. In so doing, EF-Tu increases the rate and fidelity of the translation mechanism. Recent evidence suggests that EF-Ts, the guanosine nucleotide exchange factor for EF-Tu, directly accelerates both the formation and dissociation of the EF-Tu-GTP-Phe-tRNAPhe ternary complex (Burnett, B. J., Altman, R. B., Ferrao, R., Alejo, J. L., Kaur, N., Kanji, J., and Blanchard, S. C. (2013) J. Biol. Chem. 288, 13917–13928). A central feature of this model is the existence of a quaternary complex of EF-Tu/Ts·GTP·aa-tRNAaa. Here, through comparative investigations of phenylalanyl, methionyl, and arginyl ternary complexes, and the development of a strategy to monitor their formation and decay using fluorescence resonance energy transfer, we reveal the generality of this newly described EF-Ts function and the first direct evidence of the transient quaternary complex species. These findings suggest that EF-Ts may regulate ternary complex abundance in the cell through mechanisms that are distinct from its guanosine nucleotide exchange factor functions.

Guanosine nucleotide-binding proteins (G-proteins) regulate a diverse range of essential cellular processes in all three kingdoms of life, including the universally conserved and energy-intensive process of protein synthesis. Elongation factor Tu (EF-Tu) 2 in bacteria and elongation factor 1A (eEF1A) in eukaryotes are conserved members of the TRAFAC (translation factor) class of G-proteins comprising three globular domains (I-III). Both proteins play a critical role in translation by chaperoning messenger RNA (mRNA) codon-dependent incorporation of aminoacyl-tRNA (aa-tRNA) into the aminoacyl (A) site of the two-subunit ribosome during the substrate selection step of the elongation cycle (2)(3)(4)(5)(6).
EF-Tu, through its capacity to form a ternary complex with aa-tRNA in the presence of guanosine triphosphate (GTP) (EF-Tu⅐GTP⅐aa-tRNA), increases both the rate and fidelity of aa-tRNA selection on the ribosome (7-10) (see Fig. 1A). EF-Tu simultaneously promotes ribosome binding and couples the aa-tRNA selection mechanism to irreversible GTP hydrolysis (11). Following the release of inorganic phosphate (12), aa-tRNA dissociates from EF-Tu⅐guanosine diphosphate (GDP) to enter the ribosome where it undergoes irreversible peptide bond formation (13,14).
The rates of protein synthesis and cellular growth critically depend on and are highly correlated with the availability of the complete complement of ternary complex species needed to translate the mRNA open reading frame (15)(16)(17)(18)(19). In Escherichia coli, there are ϳ86 genes encoding for elongator tRNAs that exhibit a range of expression levels (19 -21). Depending on the species, organism, and cellular growth rate, the total tRNA concentration in the cell ranges from 100 M to 1 mM (22). Although a component of these fluctuations likely relates to the proximity of each gene to the origin of replication as postulated for ribosomal RNA (20,24), changes in tRNA gene expression also reflect regulatory cues (21).
Ternary complex levels also depend on the extent to which tRNAs are aminoacylated, which is a prerequisite for EF-Tu⅐GTP binding (7,25). Direct measurements suggest that ϳ80% of expressed tRNAs are charged during logarithmic growth in nutrient-rich conditions (26,27). Under nutrientpoor conditions, however, where amino acid reserves become limiting, this level drops substantially (27,28). Such conditions may ultimately trigger the stringent response leading to the down-regulation of the translational machinery (27).
The abundance of distinct ternary complex species is also dictated by the affinity of EF-Tu for each aa-tRNA. EF-Tu exhibits tRNA-specific dissociation constants ranging from a few nanomolar to a few hundred nanomolar (8,29,30). A sub-stantial proportion of the overall binding energy between EF-Tu⅐GTP and aa-tRNA arises from contacts between the aminoacyl moiety of tRNA and the amino acid binding pocket formed at the interface of domains I and II of EF-Tu (31-33) (see Fig. 1B). In the case of the phenylalanyl-tRNA Phe ternary complex (Phe-TC), the aromatic ring of the 2Ј-linked phenylalanyl moiety has been shown to directly interact with the conserved histidine 66 (His-66) residue of EF-Tu. The switch regions of EF-Tu (S1 and S2), which are present in all G-proteins and facilitate nucleotide interactions, help position His-66, thereby enabling EF-Tu to read out the aminoacyl state of the tRNA molecule in a manner that depends on the phosphorylation state of the guanosine nucleotide (32, 34) (see Fig. 1B).
The distinct chemical and physical properties of each amino acid predict significant differences in EF-Tu binding affinity to distinct tRNA species (31,32). These disparities are thought to be dampened by a thermodynamic compensation mechanism based on unique EF-Tu interactions with sequence determinants in the tRNA acceptor stem and T-loop ( Fig. 1C) that result in a relatively uniform EF-Tu⅐GTP affinity for each aa-tRNA species (29,35,36). In this model, the tRNA species bearing an aminoacyl moiety that contributes strongly to the free energy of ternary complex affinity have tRNA bodies that bind weakly to EF-Tu. Conversely, tRNAs bearing aminoacyl moieties that contribute relatively little to the free energy of ternary complex formation bind strongly to EF-Tu (31,36). Taken together with structural and kinetic evidence that EF-Tu undergoes large scale conformational changes during aa-tRNA binding and release (34,(37)(38)(39), this model posits a degree of coupling between these two distinct determinants of EF-Tu affinity. Under steady-state, logarithmic growth conditions where aa-tRNA, EF-Tu, and GTP are at high concentrations, the vast majority of aa-tRNAs are bound to EF-Tu⅐GTP in ternary complex (40). At growth saturation or under nutrient-limited conditions, however, where the effective aa-tRNA concentrations are reduced, lower affinity species shift away from the EF-Tu⅐GTP-bound state, reducing their cellular concentrations.
Finally, the total ternary complex concentration in the cell is proportional to the concentration of GTP and the activated EF-Tu⅐GTP pool. The concentration of this latter complex in turn depends on the function of elongation factor Ts (EF-Ts), the guanosine nucleotide exchange factor for EF-Tu. The guanosine nucleotide exchange factor functions of EF-Ts facilitate the conversion of EF-Tu⅐GDP to EF-Tu⅐GTP following each round of tRNA selection. Given that the amount of EF-Tu⅐GDP is proportional to the concentration of ribosomes in the cell (10 -100 M) and the rate of protein synthesis (approximately 2-20 amino acids polymerized per second), high concentrations of EF-Ts promote a rapid rate of cell growth (41). In E. coli, the concentration of EF-Ts is tightly coupled with ribosome biogenesis and thus the ribosome concentration (15,(41)(42)(43)(45)(46)(47). Hence, a stoichiometry of EF-Tu/Ts⅐GDP in the cell that approaches 1:1 with ribosomes would facilitate prompt nucleotide exchange on EF-Tu and the maintenance of the highest concentration of EF-Tu⅐GTP possible.
Our recent efforts, which established a kinetic framework for examining ternary complex stability using fluorescence-based methods (1), provided evidence that EF-Ts directly regulates conformational changes in EF-Tu that facilitate the formation and disassembly of the Phe-TC. These observations suggested that EF-Ts has a capacity to directly regulate ternary complex concentration via the formation of an EF-Tu/Ts⅐GTP⅐aa-tRNA quaternary complex (1). Here, we directly examined whether such findings are general in nature by measuring the impact of EF-Ts on the kinetic properties of distinct ternary complex species that exhibit distinct apparent K D values (29). These investigations, combined with fluorescence resonance energy transfer (FRET) methods that facilitated the direct monitoring of EF-Tu/Ts interactions, yielded a general kinetic framework that explains how EF-Ts may influence the thermodynamic compensation model underpinning ternary complex stability, why quaternary complex has escaped detection previously, and , and the anticodon loop (green) as well as GDPNP (carbon atoms in gray, nitrogen in blue, oxygen in red, and phosphorus in orange) and the acp 3 U47 residue, the site of fluorophore labeling used in this report. B, the aminoacyl binding pocket is stabilized by the GTP-dependent coordination of switch 1 (S1; red) and switch 2 (S2; orange) regions. C, conserved contacts between the acceptor stem of aa-tRNA and all three domains of EF-Tu. The approximate position of the C terminus (CTerm) of EF-Tu is shown.
how cellular interactions between EF-Ts and ternary complex can influence the regulation of protein synthesis in the cell.

EXPERIMENTAL PROCEDURES
Purification of Native Elongation Factors-Cleavable His 6 -EF-Tu (tufB) and His 6 -EF-Ts (tsf) were expressed recombinantly in DH5␣ cells and purified by nickel-nitrilotriacetic acid affinity chromatography in the absence of magnesium as described previously (48). After removal of His 6 , the elongation factors were purified further using Superdex 75 gel filtration chromatography as described previously (1).
Purification and Labeling of Elongation Factors-Labeled EF-Tu was constructed and purified using the expression system described for native EF-Tu with the modification of a C-terminal acyl-carrier protein recognition epitope (AcpS). Labeling was achieved by incubating purified EF-Tu (C-terminal AcpS) with a 10-fold excess of CoA-linked cyanine fluorophore (either Cy3 or Cy5Q) and a 5-fold excess of AcpS phosphopantetheinyltransferase enzyme in buffer A (50 mM HEPES, pH 7.5, 10 mM Mg(OAc) 2 ) for 2 h. EF-Ts (⌬C, A160C) was generated by site-directed mutagenesis of its two native cysteines (at positions 23 and 78) to serines and Ala-160 to cysteine. EF-Ts (⌬C, A160C) was labeled by incubation with a 2-fold excess of Cy5 maleimide for 90 min at room temperature in buffer B (100 mM phosphate-buffered saline, pH 6.5, 100 M tris(2-carboxyethyl)phosphine, 100 mM NH 4 Cl).
Purification and Labeling of tRNA-Wild-type E. coli tRNA Phe was expressed and purified from the bacterial strain MRE600 as described previously (49). tRNA Arg and tRNA Met were purchased from Chemical Block and further purified by hydrophobic interaction chromatography (using a Tosoh Corp. Phenyl-5PW column) by applying a shallow gradient from buffer C (10 mM NH 4 OAc, pH 5.8, 1.7 M (NH 4 ) 2 SO 4 ) to buffer D (10 mM NH 4 OAc, pH 5.8, 10% methanol) over 20 column volumes. tRNA molecules were site-specifically labeled with either Cy3 or Cy3B via the naturally occurring modified nucleotide acp 3 U at position 47 (48,50). This labeling strategy has been shown previously to not inhibit aminoacylation or translation reactions (48,50). Each tRNA was aminoacylated by adding a 0.1ϫ molar ratio of tRNA synthetase and a 1000-fold excess of amino acid in a volume of 10 l in buffer E (50 mM Tris-HCl, pH 8, 20 mM KCl, 100 mM NH 4 Cl, 1 mM DTT, 2.5 mM ATP, 0.5 mM EDTA, 10 mM MgCl 2 ) and incubating at 37°C for 20 min. The extent of aminoacylation was verified by analytical hydrophobic interaction chromatography.
Fluorometer Experiments-The fluorescence measurements shown in Figs. 2, 4, and 5 were obtained using a Photon Technology International fluorescence spectrometer with a 532-nm long pass emission filter (LP03-532RS-25 RazorEdge by Semrock). Reactions were analyzed in buffer F (100 mM HEPES, pH 7, 20 mM KCl, 100 mM NH 4 Cl, 1 mM DTT, 0.5 mM EDTA, 2.5 mM MgCl 2 , 10 M GTP) in a 3-ml quartz cuvette with constant mixing at room temperature. Measurements were made by manually adding EF-Tu or an EF-Tu/Ts complex (preincubated in buffer F with 10 M GTP for 10 min at room temperature) to a reaction of 5 nM fluorescently labeled aa-tRNA while exciting at 532 nm and monitoring at 565 nm. The dissociation constant (K D ) was determined as described pre-viously (1). Apparent rates (k app ) were determined from presteady-state measurements where the time-dependent changes in fluorescence observed were fit to a single exponential function (51).
Rapid Stopped-flow Experiments-The fluorescence measurements shown in Figs. 3, 6, and 7 were obtained in buffer F at room temperature using an SX20 stopped-flow spectrometer from Applied Photophysics equipped with either a 550-nm long pass emission filter (OG550 by Schott) or a 633-nm long pass emission filter (LP01-633RS-25 RazorEdge by Semrock) for monitoring Cy3 and Cy5 fluorescence, respectively. All concentrations referenced herein are stated as final concentrations after mixing. Error bars represent S.E. of three separate reactions.
Nucleotide Purification-GTP and GDP were purchased from Sigma and further purified on a Tricorn Mono Q 5/50 GL ion exchange column as described previously (1).
Simulation of Ternary Complex Formation-We simulated the formation of ternary complex using MATLAB (R2010b) built on the following six reactions.

EF-Tu/Ts
The simulation was conducted with the ode45 (Dormand-Prince) solver with an absolute tolerance setting of 1.0⅐10 Ϫ6 using the starting conditions of 2 M EF-Tu/Ts, 1 mM GTP, and 250 nM aa-tRNA.

EF-Tu Binds Distinct aa-tRNAs with High
Affinity-To monitor ternary complex formation, we used a steady-state, fluorescence-based approach that yields an increase in relative intensity of Cy3 fluorescence when EF-Tu binds to site-specifically labeled aa-tRNA (1). We measured the apparent affinity of highly purified, nucleotide-free EF-Tu or an EF-Tu/Ts complex for three distinct elongator tRNA species, each containing the same acp 3 U post-transcriptional modification at position 47 used for Cy3 labeling: Phe-tRNA Phe , Met-tRNA Met , and Arg-tRNA Arg . For these experiments, EF-Tu and EF-Tu/Ts were preincubated with an excess of GTP (10 M) to ensure complex formation ("Experimental Procedures"). In each case, EF-Tu⅐ GTP or EF-Tu/Ts⅐GTP was titrated into a 1.5-ml reaction containing 5 nM aa-tRNA (Cy3-acp 3 U) and 10 M GTP ("Experimental Procedures"). As anticipated from previous studies on Phe-tRNA Phe (Cy3-acp 3 U) (1), each aa-tRNA species exhibited a marked increase in Cy3 relative fluorescence as a function of increasing factor concentration (Fig. 2). Under the current experimental conditions ("Experimental Procedures"), Cy3 fluorescence increased ϳ32, 45, and 30% for Phe-tRNA Phe , Met-tRNA Met , and Arg-tRNA Arg , respectively. As expected, no change in fluorescence intensity was observed when identical experiments were performed either in the absence of GTP or when the tRNA was not aminoacylated. To provide an estimate of the affinity of each factor for aa-tRNA from these experiments, each titration was fit to an equation for a bimolecular interaction (1). As expected from prior investigations (7,29), all three aa-tRNA species exhibited an apparent K D in the nanomolar range both in the presence and absence of EF-Ts ( Fig. 2 and Table 1). The lowest apparent K D was for Phe-tRNA Phe (6 nM), whereas Arg-tRNA Arg was the highest (22 nM) ( Table 1). Compared with the relatively large impact on Phe-TC affinity (a 3-fold increase) in the presence of EF-Ts (1), the apparent K D values of Met-tRNA Met and Arg-tRNA Arg were only modestly affected (approximately Ϯ20 -30%; Table 1).

EF-Ts Accelerates the Rate of Ternary Complex Formation-
To delineate the kinetic features governing ternary complex formation, pre-steady-state, rapid mixing experiments were performed using the same Cy3-labeled Phe-tRNA Phe , Met-tRNA Met , and Arg-tRNA Arg reagents under conditions estimating the physiological concentrations of each component (5 M factor, 1 mM GTP, and 100 nM aa-tRNA). Purified EF-Tu and EF-Tu/Ts complex were pre-incubated with saturating concentrations of GTP (1 mM) to ensure complete nucleotide binding. For all three aa-tRNAs, the apparent rate of ternary complex formation using EF-Tu was between ϳ8 and ϳ14 s Ϫ1 (Fig. 3, open bars). Identical experiments conducted using the EF-Tu/Ts complex resulted in an increase in the apparent rate of ternary complex formation for each tRNA species to ϳ40 to ϳ70 s Ϫ1 (Fig. 3, solid bars). As suggested for Phe-TC formation (1), these data argue that EF-Ts also facilitates rate-determining conformational events in EF-Tu that are required for nucleotide-dependent binding to both Met-tR-NA Met and Arg-tRNA Arg .
EF-Ts Facilitates Ternary Complex Dissociation in the Presence of GDP-To examine whether EF-Ts influences the dissociation pathways of these two distinct ternary complex species as has been demonstrated for the Phe-TC (1), we performed GDP chase experiments on Met-TC and Arg-TC preformed with either EF-Tu or EF-Tu/Ts. As noted previously (1), a hallmark kinetic signature of the EF-Tu/Ts⅐GTP⅐aa-tRNA quater-

Direct Evidence of EF-Tu/Ts⅐GTP⅐aa-tRNA Quaternary Complex
nary complex is the factor concentration dependence of the ternary complex dissociation rate in the presence of GDP. In such experiments, ternary complex is formed by adding 0.4 M factor (preincubated with 10 M GTP) to 5 nM Cy3-labeled Phe-tRNA Phe , Met-tRNA Met , or Arg-tRNA Arg . A 10-fold excess of GDP (100 M) is then added, and the rate of fluorescence decay is then tracked over time.
In the absence of EF-Ts, all three ternary complexes were observed to decay at a slow rate of ϳ0.01 s Ϫ1 , consistent with ternary complex being a stable, high affinity species (Fig. 4A). As shown previously for Phe-TC (1), the measured decay rate was independent of EF-Tu concentration as expected for a spontaneous dissociation process followed by GDP binding, an event that prevents ternary complex from re-forming (Fig. 4B). When the same experiment was performed with EF-Tu/Ts, however, the rate of fluorescence intensity decay was ϳ25-, ϳ13-, and ϳ20-fold faster for Phe-TC, Met-TC, and Arg-TC, respectively (Fig. 4A). The observed rate of decay was also linearly dependent on the concentration of EF-Tu/Ts (Fig. 4B). This trend is inconsistent with a simple unimolecular dissociation pathway, suggesting that EF-Ts facilitates ternary complex dissociation through a bimolecular interaction that influences the nucleotide binding pocket of EF-Tu. Here, the apparent EF-Ts on-rate estimated from the slope of the concentrationdependent process ranged from ϳ0.3 to 1.0 M Ϫ1 s Ϫ1 depending on the tRNA species and taking into account that the concentration of free EF-Ts is roughly 50% of the total factor concentration. As this rate is more than 2 orders of magnitude slower than a diffusion-limited process as well as the established rate of nucleotide binding to EF-Tu (52), we conclude that one or more rate-determining events in EF-Tu, likely related to GTP release, precede the EF-Ts-facilitated ternary complex dissociation process leading to the loss of fluorescence.

EF-Ts Increases the Rate of Ternary Complex Turnover under
Steady State Conditions-To probe whether EF-Ts influences rate-determining step that precede the dissociation process observed in the presence of GDP, we next sought to ascertain whether the rate of ternary complex turnover was influenced by EF-Ts under steady-state conditions. To do so, we developed a FRET-based approach to directly examine the rate of ternary complex turnover by site-specifically labeling EF-Tu with a quencher fluorophore at its C terminus ("Experimental Procedures"). The C terminus of EF-Tu and the acp 3 U47 residue in tRNA are in close proximity (approximately 30 Å) (Fig.  1). To dampen the environment-dependent increase in Cy3 fluorescence that accompanies ternary complex formation (Fig.  2), Phe-tRNA Phe , Met-tRNA Met , and Arg-tRNA Arg were labeled at their acp 3 U47 residues using Cy3B, a fluorophore less prone to cis-trans isomerization and environment-dependent changes in quantum yield (53)(54)(55). Hence, the formation of ternary complex is anticipated to result in a substantial decrease in Cy3B fluorescence intensity as a result of energy transfer to the nearby quencher (Fig. 5A).
Using this approach, ternary complex formation resulted in an ϳ65% decrease in Cy3B fluorescence intensity for all three aa-tRNAs examined (Fig. 5B). As expected, this change in signal was strictly dependent on GTP and the aminoacyl moiety of aa-tRNA (Fig. 5B). Consistent with the findings presented for unlabeled EF-Tu (Fig. 4), this reduction in Cy3B fluorescence intensity could be fully recovered by the addition of a 10-fold molar excess of GDP (data not shown). These data convincingly revealed that quencher-labeled EF-Tu is fully competent to form ternary complex.
To directly examine ternary complex turnover under steadystate conditions, a competition experiment was performed in which a 10-fold molar excess of unlabeled EF-Tu or EF-Tu/Ts was added to a preformed, quencher-labeled EF-Tu⅐GTP⅐Phe-tRNA Phe (Cy3B-acp 3 U47) ternary complex. As the vast majority of ternary complex re-formed after dissociation will not be quenched, the intrinsic rate of ternary complex turnover in this assay is revealed by the rate of increase in Cy3B fluorescence intensity.
In both the absence and presence of EF-Ts, Cy3B fluorescence recovered at a rate that was tRNA-specific (Fig. 5, C-F). For the unlabeled EF-Tu chase experiment, the rates of steadystate ternary complex turnover approximated those observed upon quenching with GDP in the absence of EF-Ts (Fig. 5C and compare Table 2 k off with Table 1 k app,2 ). However, distinct responses were evidenced in each EF-Tu/Ts chase experiment (Fig. 5, D-F). Here, the steady-state turnover rates were approximately an order of magnitude slower than those observed in the presence of GDP (compare Table 1 k app,2 with Table 2 k off ).
These findings suggest that EF-Ts has the propensity to destabilize each ternary complex and to specifically promote ternary complex dissociation in the presence of GDP. Such trends are consistent with the notion that nucleotide exchange occurs at a slow rate within ternary complex and that EF-Ts increases the rate of such processes via direct interactions with EF-Tu while it remains bound to aa-tRNA (1). We speculate that, when only GTP is present in the reaction, nucleotide can reload onto EF-Tu prior to its dissociation from aa-tRNA. When GDP is present in excess of GTP, however, GDP binding to EF-Tu leads to ternary complex dissociation. We therefore infer that EF-Ts can act as a guanosine nucleotide exchange factor for EF-Tu while it is bound to aa-tRNA via a quaternary complex where the formation of an EF-Tu⅐GDP⅐aa-tRNA ter-nary complex stimulates EF-Tu release from aa-tRNA, which is accelerated by EF-Ts.

EF-Tu/Ts⅐GTP Is an Abundant Species under Cellular
Conditions-The notion that EF-Ts can interact with EF-Tu while it is bound to aa-tRNA provides a plausible model for the observation that EF-Ts accelerates the process of ternary com-  plex formation (Fig. 3). As speculated previously (1), EF-Ts may increase the rate of ternary complex formation via a transient EF-Tu/Ts⅐GTP⅐aa-tRNA quaternary complex. A key issue with this model is that the EF-Tu/Ts⅐GTP species is expected to dissociate at a rate of ϳ60 s Ϫ1 at physiological GTP concentrations (approximately 1 mM) (52). Meaningful concentrations of the EF-Tu/Ts⅐GTP complex, however, are expected at EF-Ts concentrations above 1 M as the bimolecular rate constant between EF-Ts and EF-Tu⅐GTP is 3⅐10 7 M Ϫ1 s Ϫ1 (52). Indeed, such considerations predict that ϳ50% of EF-Tu will be in an EF-Tu/Ts⅐GTP complex under the conditions where EF-Ts is observed to accelerate ternary complex formation (Fig. 3). At cellular EF-Ts concentrations, which under nutrient-rich conditions are estimated to be Ն10 M (41,42), the proportion of EF-Tu/Ts⅐GTP complex is expected to be even higher as the apparent rate of EF-Tu/Ts⅐GTP re-formation will be as much as ϳ5-fold faster than the rate of EF-Ts dissociation from EF-Tu⅐GTP.
To directly test these predictions, we set out to measure the interactions between EF-Tu and EF-Ts through a FRET-based approach (Fig. 6A). To do so, we used enzymatic tagging procedures to site-specifically label EF-Tu with a Cy3 fluorophore at its C terminus and EF-Ts with a Cy5 fluorophore through maleimide chemistry at a position engineered to yield high FRET (A160C) (38) ("Experimental Procedures"). When 2 M Cy3-labeled EF-Tu was rapidly mixed with 2 M Cy5-labeled EF-Ts under stopped flow ("Experimental Procedures"), a rapid increase in Cy5 emission was observed (k app ϭ 18.5 s Ϫ1 ), which as expected from the kinetics and high affinity nature of the EF-Tu/Ts interaction (52) remained stable over time (Fig. 6, A and B, black curve, and Table 3).
To determine the dissociation rate of the EF-Tu/Ts complex in the presence of saturating concentrations of GTP, a preformed EF-Tu(Cy3)/Ts(Cy5) complex was rapidly mixed with 1 mM GTP in the presence of a 10-fold molar excess of unlabeled EF-Ts to prevent the complex from re-forming (Fig. 6, A and B, blue curve). As expected (52), a rapid loss of fluorescence was observed (ϳ148 s Ϫ1 ) ( Table 3). This value is ϳ2-fold faster that the EF-Tu/Ts⅐GTP dissociation rate estimated using native components (52), suggesting that the modifications to EF-Tu and EF-Ts may modestly destabilize the EF-Tu/Ts complex.
We next investigated the extent to which GTP disrupts the EF-Tu/Ts complex under steady state conditions in the absence of unlabeled EF-Ts where a dynamic equilibrium between dissociation and association is expected. In line with known kinetic parameters of this system (52), mixing the preformed EF-Tu(Cy3)/Ts(Cy5) complex (2 M) with GTP (1 mM) led to a rapid loss of FRET, which stabilized at approximately ϳ55% of the starting intensity (Fig. 6, A and B, red curve, and Table 3). These data argue that the EF-Tu/Ts⅐GTP complex is an abundant species both in our experiments and the cellular milieu.
To test the hypothesis that EF-Ts accelerates the binding of aa-tRNA to EF-Tu⅐GTP, a preformed EF-Tu(Cy3)/Ts(Cy5) complex was rapidly mixed with 1 mM GTP and 500 nM Phe-tRNA Phe (Fig. 6, A and B, cyan curve). Here, Cy5 fluorescence was observed to decay at a rate of ϳ242 s Ϫ1 (Table 3), a rate significantly faster than when only GTP or unlabeled EF-Ts was added (Fig. 6B, inset). This finding suggests that aa-tRNA contributes to the apparent rate of EF-Tu(Cy3)/Ts(Cy5) dissociation as would be expected if ternary complex is formed rapidly (approximately Ն140 s Ϫ1 ).  Quaternary Complex Is Transient by Nature-To directly observe the EF-Tu/Ts⅐GTP⅐aa-tRNA quaternary complex, we designed an experiment to detect FRET between EF-Ts and aa-tRNA during ternary complex formation where the hypothesized architecture of quaternary complex predicts a physical proximity between the Cy3B fluorophore at position U47 of aa-tRNA and Cy5 at position 160 of EF-Ts (1) (Fig. 7A). The conditions of this experiment were derived by simulating the expected rate of quaternary complex formation and decay that should accompany ternary complex formation based on a framework that included the known kinetic features of the interactions among EF-Tu, EF-Ts, guanosine nucleotide, and Phe-tRNA Phe (1, 52) (Fig. 7B, red curve). Using this approach, the optimal conditions for detecting quaternary complex were determined to be a concentration of 2 M EF-Tu/Ts, 250 nM Phe-tRNA Phe , and 1 mM GTP.
As anticipated by these simulations, when 2 M EF-Tu/ Ts(Cy5) was rapidly mixed with 250 nM Phe-tRNA Phe (Cy3Bacp 3 U47) and 1 mM GTP, FRET was observed between the excited Cy3B fluorophore and the Cy5-labeled EF-Ts (Fig. 7C,  black). No evidence of FRET was observed when the exact same experiment was performed using deacylated tRNA Phe (Fig. 7C,  gray).
The biphasic nature of the transient FRET signal observed was fit to a double exponential process to estimate the apparent rates of quaternary complex formation and decay, respectively. Here, the rise and decay portions were estimated to occur at rates of ϳ870 Ϯ 23 and 143 Ϯ 7 s Ϫ1 , respectively (Table 4). These apparent rates suggest that the appearance of FRET is diffusion-limited and that the rate of quaternary complex decay and thus ternary complex formation may be as fast as 140 s Ϫ1 . Notably, this rate is strikingly consistent with our stopped-flow measurements based on changes in Cy3 relative fluorescence intensity where ternary complex formation occurs at a rate of   ϳ70 s Ϫ1 (Fig. 3) under conditions in which ϳ50% of EF-Tu is expected to be in an EF-Tu/Ts⅐GTP complex (Fig. 6). Such findings provide compelling, direct evidence that ternary complex formation occurs via a highly transient quaternary complex species under cellular conditions.

DISCUSSION
Contemporary models posit that the EF-Tu life cycle parallels the quintessential model of G-protein function (56 -58). After aa-tRNA is delivered to the actively translating ribosome, EF-Tu⅐GDP is released as an inactive species whereupon the guanosine nucleotide exchange factor, EF-Ts, operates on it to exchange GDP for GTP. Thereafter, EF-Tu⅐GTP is released as an activated species that is again competent to bind free aa-tRNA with uniformly high affinity (13,14,36,52,(57)(58)(59)(60).
The data presented here and elsewhere (1,61) argue that refinements to this model are needed. At cellular EF-Ts concentrations (approximately Ͼ1 M), our findings suggest that ternary complex preferentially forms via a transient EF-Tu/ Ts⅐GTP⅐aa-tRNA quaternary complex (Fig. 8). This pathway is kinetically preferred because EF-Ts lowers the activation barrier for rate-determining conformational processes in EF-Tu that accompany the formation of a high affinity complex with aa-tRNA. Given what is presently known, the nature of these conformational processes likely surrounds the allosteric linkage between the aminoacyl moiety linked to the terminus of aa-tRNA and the switch regions surrounding the GTP nucleotide that coordinate magnesium and the ␥-phosphate group (Fig. 1).
Our direct measurements suggest that the lifetime quaternary complex is on the order of ϳ10 ms (Fig. 7). As the predominant means for investigating ternary complex formation have entailed size exclusion chromatography, filter binding, and gel mobility shift assays (8,23,62) where rapid kinetic studies have been specifically lacking, its inherent instability is likely the key reason why quaternary complex has escaped detection previously. Although the existence of a quaternary encounter complex species is an expected result based on purely physical considerations, the capacity of EF-Ts to increase the rate of ternary complex formation substantially above that of protein synthesis (approximately 100 versus 10 -20 s Ϫ1 ; Figs. 3 and 7) argues that it serves a critical biological role related to replenishing the supply of ternary complexes as they are consumed by the process of protein synthesis.
The data presented also suggest that EF-Ts destabilizes ternary complex species to make their abundance sensitive to the nucleotide status of the cell (Fig. 5). Such tendencies are consistent with EF-Ts remodeling the switch regions in EF-Tu while it is bound to aa-tRNA (as for the free protein (44)) to influence the rate of nucleotide exchange. Two notable corollaries to this observation are that 1) the nucleotide exchange rate may not be the same for all tRNA species, and 2) the activated EF-Tu molecule in ternary complex may be deactivated by energy-neutral processes. Evidence supporting the first corollary is shown for the three tRNA species investigated in this study (Phe-tRNA Phe , Met-tRNA Met , and Arg-tRNA Arg ) where the effects of EF-Ts on ternary complex differ by as much as an order of magnitude (Figs. 3, 4, and 5).
As the effective concentration of quaternary complex will be strongly dependent on EF-Ts concentration, such effects are also expected to be highly responsive to the active EF-Ts pool. Although the second corollary is likely irrelevant under nutrient-rich conditions where the intracellular concentration of GTP is much higher than GDP, it may play a significant role in regulating ternary complex abundance upon a change in the GTP/GDP ratio. EF-Ts rapidly dissociates ternary complex when GDP loads onto EF-Tu while it is bound to TC (Figs. 4, 5, and 8). Correspondingly, changes in EF-Ts concentration as well as the GDP/GTP ratio are likely to contribute differentially to the relative stabilities and effective concentrations of specific ternary complex species in interdependent ways. Although it is not presently clear whether EF-Ts contributes evenly or differentially to the stabilities of distinct ternary complex species, the data obtained for the three ternary complex species assessed here suggest that ternary complex affinities become somewhat more dispersed in the presence of EF-Ts (Table 1). Such differences likely arise from a balance of the distinct contributions to ternary complex affinity, including contacts between aa-tRNA and EF-Tu as well as the nucleotide exchange rate. A key prediction of these considerations is that the abundance and lifetime of each ternary complex type will be sensitive to the metabolic state of the cell where EF-Ts can serve to rapidly dissipate the ternary complex pool through an energy-neutral mechanism in the event of cellular stress. Such a capacity would provide a means for the cell to immediately cease the energy-expensive process of translation without waiting for ternary complex to be consumed and allow for the repurposing of GTP. The demonstration that EF-Ts actively resolves quaternary complex species that contain GDP (Fig. 5) also has potentially important implications for the mechanism of tRNA selection as an EF-Tu⅐GDP⅐aa-tRNA ternary complex is formed each time an actively translating ribosome decodes mRNA (59). A key open question that must be resolved through future investigations is whether EF-Ts is able to gain access to the EF-Tu⅐GDP⅐aa-tRNA ternary complex while engaged at the A site of the ribosome. If so, EF-Ts may accelerate the rate of EF-Tu⅐GDP release from the ribosome following aa-tRNA entry into the peptidyltransferase center, a step reported to be rate-limiting to the process (13,14). To our knowledge such questions have yet to be explored.