Mg2+ is not catalytically required in the intrinsic and kirromycin-stimulated GTPase action of Thermus thermophilus EF-Tu.

The influence of divalent metal ions on the intrinsic and kirromycin-stimulated GTPase activity in the absence of programmed ribosomes and on nucleotide binding affinity of elongation factor Tu (EF-Tu) from Thermus thermophilus prepared as the nucleotide- and Mg(2+)-free protein has been investigated. The intrinsic GTPase activity under single turnover conditions varied according to the series: Mn(2+) (0.069 min(-1)) > Mg(2+) (0.037 min(-1)) approximately no Me(2+) (0.034 min(-1)) > VO(2+) (0.014 min(-1)). The kirromycin-stimulated activity showed a parallel variation. Under multiple turnover conditions (GTP/EF-Tu ratio of 10:1), Mg(2+) retarded the rate of hydrolysis in comparison to that in the absence of divalent metal ions, an effect ascribed to kinetics of nucleotide exchange. In the absence of added divalent metal ions, GDP and GTP were bound with equal affinity (K(d) approximately 10(-7) m). In the presence of added divalent metal ions, GDP affinity increased by up to two orders of magnitude according to the series: no Me(2+) < VO(2+) < Mn(2+) approximately Mg(2+) whereas the binding affinity of GTP increased by one order of magnitude: no Me(2+) < Mg(2+) < VO(2+) < Mn(2+). Estimates of equilibrium (dissociation) binding constants for GDP and GTP by EF-Tu on the basis of Scatchard plot analysis, together with thermodynamic data for hydrolysis of triphosphate nucleotides (Phillips, R. C., George, P., and Rutman, R. J. (1969) J. Biol. Chem. 244, 3330-3342), showed that divalent metal ions stabilize the EF-Tu.Me(2+).GDP complex over the protein-free Me(2+).GDP complex in solution, with the effect greatest in the presence of Mg(2+) by approximately 10 kJ/mol. These combined results show that Mg(2+) is not a catalytically obligatory cofactor in intrinsic and kirromycin-stimulated GTPase action of EF-Tu in the absence of programmed ribosomes, which highlights the differential role of Mg(2+) in EF-Tu function.

The GTPase superfamily of proteins, known more commonly as G-proteins, are ubiquitous in cellular systems and serve as key regulatory molecules catalyzing the hydrolysis of the ␤,␥phosphate bond in GTP (1)(2)(3)(4). For most G-proteins this process yields a tightly bound enzyme-product complex requiring the action of a nucleotide exchange factor to kinetically facilitate exchange of the GTP for GDP in the active site. Nucleotide binding and hydrolysis are key to regulating the activity of these enzymes, and both of these steps occur in the presence of the divalent metal ion Mg 2ϩ as the naturally occurring cofactor in the cell. However, in EF-Tu 1 function it has not been possible hitherto to establish whether there is a differential role of the divalent metal ion in nucleotide hydrolysis as opposed to nucleotide binding.
In all G-proteins defined hitherto by x-ray crystallographic studies, the substrate analog guanosine 5Ј-(␤,␥-imido)-triphosphate is bound as a complex with Mg 2ϩ in which the divalent metal ion is coordinated to oxygen atoms from the ␤-phosphate and ␥-phosphate groups, and the guanine and divalent metal ion binding sites appear to be tightly coupled (5,6). For ras (7), for elongation factor Tu from both thermophilic bacteria (8 -11) and Escherichia coli (12,13), and for G␣-subunits of heterotrimeric G-proteins (14,15), the dissociation constants for release of protein-bound nucleotide measured in the presence of Mg 2ϩ indicate differential binding affinities for GDP and GTP by at least an order of magnitude. Binding affinities for both nucleotides are often measured kinetically by ligand displacement because G-proteins are less stable when prepared in their nucleotide-free form. Because nucleotide separation is necessary for complete removal of Mg 2ϩ from the protein, the less stable, nucleotide-free form of G-proteins complicates efforts to determine whether the divalent metal ion has differential roles in nucleotide hydrolysis and nucleotide binding.
In contrast to the structural instability of elongation factor Tu of mesophiles such as E. coli, the homologous thermostable protein in its nucleotide-free form is less prone to inactivation, and comparison of the kinetics of nucleotide binding shows it to be identical to EF-Tu isolated as the GDP-bound complex from the cytosol (11,16). This characteristic has allowed application of efficient biochemical purification methods for preparation of the nucleotide-free elongation factor from Thermus thermophilus (16,17) and Bacillus stearothermophilus (18,19) for x-ray structure analysis of the complex formed with an inhibitor analog of GTP in the active site (5,6). Isolation of nucleotidefree elongation factor Tu (17)(18)(19)(20) and p21 ras (7) has also allowed comparison of divalent metal ions on nucleotide binding and hydrolysis (8,19,21,22) and the investigation of nucleotide-protein interactions by magnetic resonance methods (16,(21)(22)(23)(24)(25)(26). In these studies, however, the question of whether the divalent metal ion is an absolute requirement for GTP hydrolysis has not been addressed.
In the present study we report results of kinetic and equilibrium binding studies to compare the influence of divalent metal ions on the intrinsic and kirromycin-stimulated GTPase activ-ity of EF-Tu of T. thermophilus in the absence of programmed ribosomes. We compared the influence of Mn 2ϩ and VO 2ϩ to that of Mg 2ϩ because they have been employed as paramagnetic probes of Mg 2ϩ function in thermostable EF-Tu (21, 23). Because we have been able to show that EF-Tu when depleted of nucleotide is not associated with protein-bound Mg 2ϩ , we have been able to determine not only the influence of different divalent metal ions on the biological activity of EF-Tu without interference from adventitious Mg 2ϩ but also the intrinsic GTPase activity of the Mg 2ϩ -free protein. The results lead to the conclusion that Mg 2ϩ is not a catalytically obligatory cofactor in the intrinsic and kirromycin-stimulated GTPase activity in the absence of programmed ribosomes. We also show that the binding of divalent metal ions provides the thermodynamic driving force for stabilization of the EF-Tu⅐Me 2ϩ ⅐GDP complex, the effect being greatest with Mg 2ϩ . The differential role of Mg 2ϩ in EF-Tu function is discussed further through analysis of thermodynamic relationships governing the hydrolysis of nucleoside 5Ј-triphosphates (27,28) and the kinetic requirement for a nucleotide-exchange factor in the cell. Preparation of Nucleotide-free EF-Tu-EF-Tu was obtained as the EF-Tu⅐Mg 2ϩ ⅐GDP complex from T. thermophilus (17) or from the cell paste of E. coli (JM109) engineered with the tuf1 gene of T. thermophilus (HB8) (29). Nucleotide-and Mg 2ϩ -free EF-Tu (EF-Tu f ) was isolated as described by Limmer et al. (16), as modified previously (21). For removal of GDP and Mg 2ϩ , EF-Tu was diluted 1:6 into 5 M urea buffered to pH 5.65 with 0.01 M (NH 4 ) 2 HPO 4 . The resultant mixture with a protein concentration of ϳ1.5 ϫ 10 Ϫ4 M was applied to a 1 ϫ 52 cm column of CM-Sephadex CL-6B (Sigma) and eluted with the same buffer to remove GDP. A second buffer consisting of 0.01 M (NH 4 ) 2 HPO 4 at pH 7.5, 5 M urea, and 0.5 M KCl was then applied to elute the protein.

Materials
Renaturation of the protein was carried out by dialyzing against 0.05 M KCl and 0.15 M NH 4 Cl or against 0.2 M KCl buffered with 0.05 M HEPES at pH 7.5. An Amicon YM-10 membrane (Amicon, Inc., Beverly, MA) extensively washed prior to use was employed to concentrate the protein solution. Whereas the original description for chromatography of EF-Tu and removal of Mg 2ϩ includes 0.01 M EDTA in the buffer (16), we found no difference in the results if EDTA was not present during denaturation and cation exchange chromatography of the protein. Therefore, to compare the influence of metal ions on nucleotide binding and GTPase activity, EDTA was not used during nucleotide removal to avoid the possible influence of low levels of adventitious EDTA. Use of glass containers was also avoided to prevent contamination with trace divalent metal ions. EF-Tu f prepared in this manner was stored at Ϫ20°C in the dialysis buffer containing 50% (v/v) glycerol.

Methods
Atomic Emission Spectroscopy-The Mg 2ϩ content of protein, nucleotide, kirromycin, and of all buffer solutions was determined by inductively coupled plasma atomic emission spectroscopy with a GBC Integra XMP spectrometer (BDC Scientific Equipment, Ltd., Dandenong, Victoria, Australia). The detection limit of this method is 1.0 ϫ 10 Ϫ4 g Mg 2ϩ per liter or 4.0 ϫ 10 Ϫ6 moles (g atoms) Mg 2ϩ per liter of solvent.
Determination of Nucleotide Binding Affinity-We observed previously that variability in the stoichiometry of nucleotide binding and biphasic character in Scatchard plots (30, 31) are essentially eliminated when BSA is added to the incubation mixture (21). Because nucleotide binding to BSA alone was not detected under these conditions, we concluded that BSA provided the role of a protectant neutral macro-molecule, as is observed in a variety of biochemical assays. 2 For these reasons, the use of BSA was continued in studies to determine nucleotide binding affinity of EF-Tu f . The dissociation constants and the stoichiometry of GDP and GTP binding to EF-Tu f were determined following the nitrocellulose filter binding method of Arai et al. (9,12), as described in previous studies (21). For binding affinity measurements, all stock solutions of protein, buffer, and nucleotide were treated with Chelex to remove possible trace contaminant divalent metal ions. The reaction mixture (0.0175 ml) containing 0.05 M KCl, 8.8 ϫ 10 Ϫ9 M EF-Tu f , and 1.6 ϫ 10 Ϫ5 g BSA/ml was buffered to pH 7.5 with 0.05 M HEPES. In the reaction mixture the concentration of guanosine nucleotides varied from 12.5 ϫ 10 Ϫ9 to 500 ϫ 10 Ϫ9 M with [8-3 H]GDP or [8-3 H]GTP of specific activity 1.5 ϫ 10 Ϫ6 Ci/nmol. For nucleotide binding in the presence of Mg 2ϩ or Mn 2ϩ , metal ion was added to a final concentration of 0.01 M. Variation of the order of addition of Mg 2ϩ or Mn 2ϩ with respect to the addition of nucleotide was not observed to influence binding affinity. However, this was not the case for nucleotide binding in the presence of VO 2ϩ .
To determine nucleotide binding affinity in the presence of VO 2ϩ , low concentrations of the metal ion must be used to avoid its precipitation as polymeric VO(OH) 2 at neutral pH (32). We found the optimal concentration of VO 2ϩ to be ϳ1.0 ϫ 10 Ϫ4 M, lower concentrations resulting in greater variability of the results and higher concentrations tending to caused precipitate formation. Because VO 2ϩ may form tight complexes with nucleotides of 1:1 and 1:2 metal/nucleotide stoichiometry (33,34), we investigated the influence of the order of addition of reagents to the incubation mixture. For nucleotide binding in the presence of VO 2ϩ , it was found that addition of VO 2ϩ had to follow addition of nucleotide to the reaction mixture containing EF-Tu f and BSA to observe reproducibly maximum binding affinity and 1:1 stoichiometry of binding of GDP or GTP. In binding assays with VO 2ϩ , BSA dialyzed against 0.05 M KCl containing 0.05 M HEPES at pH 7.5 was presaturated with a 4-fold molar excess of VO 2ϩ to account for the tight binding of this metal ion by BSA (35). Also the microcentrifuge tubes were pretreated with VO 2ϩ -BSA. Pretreatment of BSA with the other metal ions was not found to be necessary.
After addition of the metal ion, the reaction mixture was incubated for 15 min at ambient temperature (ϳ20°C) according to Arai et al. (9,12), and aliquots of the reaction mixture for measurements in triplicate were applied to the nitrocellulose membrane (HA 0.54 m; Millipore Corporation, Belford, MA). The filter was then washed thrice with 1 ml of an ice-cold solution of the reaction buffer containing only 0.05 M KCl and 0.05 M HEPES at pH 7.5. The protein-bound radioactivity retained by the filter was measured after air drying with the aid of a Packard Instruments Minax␤ Tri-Carb 4000 Series Liquid Scintillation Counter (Packard Instruments, Meriden, CT 06450).
Determination of GTPase Activity-Intrinsic GTPase activity of EF-Tu f was determined by adapting the method of Peter et al. (36), as previously described (21). The reaction mixture contained either 0.05 M KCl, 0.15 M NH 4 Cl, and 1.0 ϫ 10 Ϫ5 M EF-Tu f , buffered to pH 7.5 with 0.05 M HEPES or 0.2 M KCl and 1.0 ϫ 10 Ϫ5 M EF-Tu f similarly buffered with HEPES. The latter mixture was used to prevent coordination of the vanadyl ion by the ammonium ion. 3 In single turnover experiments, GTP was added so that EF-Tu f remained in slight (ϳ1.05:1.00) excess. For multiple turnover experiments, GTP and divalent metal ion were both added to a final concentration of 1.0 ϫ 10 Ϫ4 M. To determine the kirromycin-stimulated GTPase activity, the reaction mixture additionally contained 5.0 ϫ 10 Ϫ5 M kirromycin. In both single and multiple turnover experiments, the reaction was initiated by addition of [␥-32 P]GTP of specific activity 5.0 ϫ 10 Ϫ7 Ci/nmol. To quench the reaction and to obtain maximum chromatographic resolution, aliquots of the reaction mixture were removed with time and added 1:3 (v/v) to ice-cold formic acid. The quenched aliquots were then stored on ice until application for chromatography on polyethyleneimine-cellulose sheets (Machery-Nagel, Dueren, Germany). Electronic autoradiography was carried out as described (21). The data were fitted to a single exponential curve with fixed baseline offset by a nonlinear least-squares method with the aid of the program GRAFIT3.0 (Erithacus Software, Ltd., Staines, UK).

RESULTS
Removal of Mg 2ϩ -In a variety of studies, the influence of divalent metal ions on the intrinsic, kirromycin-stimulated, and physiological GTPase activity of EF-Tu has been examined for EF-Tu of both E. coli (37,38) and thermophilic (19,20,24) organisms. Also, the influence of monovalent and divalent metal ions on the intrinsic GTPase activity of p21 ras has been compared with that of EF-Tu isolated from E. coli (39). In these studies, however, the question of whether Mg 2ϩ is a catalytically obligatory cofactor for GTP hydrolysis could not be addressed directly because the purified protein was associated with significant quantities of nucleotide and Mg 2ϩ . An important motivation of the investigations reported here was to differentiate the role of Mg 2ϩ in GTP hydrolysis from that in nucleotide binding and to determine whether the divalent metal ion is an absolute requirement for GTPase action. For this reason it was necessary to ensure efficient removal of protein-bound Mg 2ϩ .
We found that Mg 2ϩ could be consistently removed with high efficiency only under partial denaturation of EF-Tu in 5 M urea at pH 5.6 followed by cation exchange chromatography. As shown in Fig. 1, the concentration of Mg 2ϩ bound as the EF-Tu⅐Mg 2ϩ ⅐GDP complex shows a linear dependence on protein concentration. However, the amount of Mg 2ϩ associated with EF-Tu f , when prepared as described under "Experimental Procedures," was consistently below the detection limit of the atomic emission instrumentation. From Fig. 1 we can conclude that there must be less than 0.026 moles of Mg 2ϩ bound per mole of protein. By similar methods we confirmed that the Mg 2ϩ content of all other solutions, including nucleotides and kirromycin, was also equivalently negligible.
Wittinghofer and Leberman (18) proposed that Mg 2ϩ is bound to EF-Tu as the Mg 2ϩ -nucleotide complex. For this reason we separated Mg 2ϩ from the EF-Tu⅐Mg 2ϩ ⅐GDP complex together with the nucleotide by partial denaturation and cation exchange chromatography. Extensive dialysis of the B. stearothermophilus EF-Tu⅐Mg 2ϩ ⅐GDP complex against nucleotideand Mg 2ϩ -free buffer results in removal of only 55% of the nucleotide (18). In this respect we note that the stoichiometry of protein-bound Mg 2ϩ in Fig. 1 for the EF-Tu⅐Mg 2ϩ ⅐GDP complex is equivalent to about 0.6 moles Mg 2ϩ per mole of EF-Tu despite similar extensive dialysis over a 24 h period. These correlative observations affirm the need to remove nucleotide by denaturation followed by cation exchange chromatography to achieve efficient depletion of protein-bound Mg 2ϩ , as described under "Experimental Procedures." Nucleotide Binding-Whereas the emphasis of our studies was to evaluate whether Mg 2ϩ is a catalytically obligatory cofactor in the intrinsic GTPase activity of EF-Tu, we first summarize results of measurements of nucleotide binding affinity in the absence and presence of divalent metal ions. In this study we compared the influence of the divalent metal ions Mn 2ϩ and VO 2ϩ to that of Mg 2ϩ . The former has been employed in studies of EF-Tu of B. stearothermophilus as a paramagnetic substitute for Mg 2ϩ (8,24) whereas the latter paramagnetic species has been found to closely mimic structural interactions of Mg 2ϩ with nucleotides (33,34).
The binding affinity of GTP and GDP to EF-Tu f in the absence and presence of added divalent metal ions was evaluated on the basis of Scatchard plots (30,31). Table I  The influence of other divalent metal ions on nucleotide binding by T. thermophilus EF-Tu has not been previously investigated. Whereas Mg 2ϩ and Mn 2ϩ show equivalent effects in the binding of GDP to EF-Tu, as seen in Table I, the affinity for GDP in the presence of VO 2ϩ is decreased by a factor of ϳ5. On the other hand, both VO 2ϩ and Mn 2ϩ facilitate tighter binding of GTP than does Mg 2ϩ . It is of interest to note that the binding affinity is not only decreased by one to two orders of magnitude in the absence of added divalent metal ion but also both nucleotides are bound with equal affinity.
GTPase Activity- Fig. 2 illustrates progress curves for EF-Tu f -catalyzed hydrolysis of GTP in the presence of Mg 2ϩ , Mn 2ϩ , and VO 2ϩ and in the absence of added divalent metal ion under single turnover conditions. The activity in the presence of Mn 2ϩ is greater than in the presence of Mg 2ϩ , which in turn is identical to the hydrolytic activity in the absence of added divalent metal ion. The intrinsic GTPase activity of EF-Tu f in the absence of divalent metal ion has not been previously reported, and the magnitude of the effect is comparable with that in the presence of Mg 2ϩ . For these reasons we took special precautions to ensure that the protein and all reagents were exhaustively depleted of adventitious Mg 2ϩ , as described under "Experimental Procedures."   Table I showed that binding of both GDP and GTP in the presence of VO 2ϩ is of greater affinity than in the absence of added divalent metal ion, we see in Fig. 2 that hydrolysis of GTP in the presence of VO 2ϩ is slower than in the absence of divalent metal ions or in the presence of Mg 2ϩ . Whereas we are not able to provide a complete explanation of this difference at present, the affinity of VO 2ϩ for inorganic triphosphate to form a 1:1 complex is at least 2 orders of magnitude greater than that of Mg 2ϩ (40). This effect is probably because of increased covalency in metalligand interactions, which may hinder hydrolysis of the triphosphate moiety in the protein-nucleotide complex. The apparent inhibitory effect of VO 2ϩ can be abolished by substitution of NH 4 ϩ for K ϩ in the reaction mixture whereas the reaction in the presence of Mg 2ϩ or Mn 2ϩ is not influenced by addition of NH 4 ϩ . Because VO 2ϩ can be coordinated by amines and other nitrogen-donor ligands 3 (41)(42)(43), the high concentration of NH 4 ϩ is likely to cause coordination to VO 2ϩ , converting the system to be equivalent to that in the absence of added divalent metal ion. 4 Table II compares the results of our studies showing the influence of divalent metal ions under single turnover conditions for both intrinsic GTPase and kirromycin-stimulated GTPase activity in the absence of programmed ribosomes. For measurements of intrinsic GTPase activity, the value of k cat of 0.037 min Ϫ1 in the presence of Mg 2ϩ is in good agreement with that of 0.042 min Ϫ1 reported by Zeidler et al. (11). On the other hand, the value of k cat for GTP hydrolysis in the presence of Mn 2ϩ is 0.069 min Ϫ1 , greater than that observed in the presence of Mg 2ϩ . The effect of Mn 2ϩ on accelerating hydrolysis of GTP over that observed with Mg 2ϩ has been observed also for the intrinsic GTPase activity of the ras proteins p21, Ran, and Rap1A (44). As seen in Table II, the kirromycin-stimulated GTPase activity in the absence of programmed ribosomes fol-lows a similar dependence on added divalent metal ion to that observed for intrinsic GTPase activity. 5 The hydrolysis of GTP catalyzed by EF-Tu f under multiple turnover conditions is illustrated in Fig. 3. It is seen that substrate turnover is faster in the presence of Mn 2ϩ than in the presence of Mg 2ϩ . On the other hand, substrate turnover is slower in the presence of Mg 2ϩ than in its absence. This latter observation affirms that significant amounts of contaminant Mg 2ϩ could not have been present in the metal-free case under multiple turnover conditions because substrate turnover is faster in the absence of added divalent metal ion. Substrate hydrolysis under multiple turnover conditions occurs because of a combination of the intrinsic GTPase action of the protein together with nucleotide exchange. The different pattern of the influence of metal ions under multiple turnover conditions, therefore, must take into account the relative catalytic effects observed under single turnover conditions, as modulated by metal-dependent variations in the on-and off-rate constants for substrate binding and product release.

Whereas Scatchard analyses summarized in
As evident in Table I, the affinity of EF-Tu f for GDP is identical in the presence of Mg 2ϩ and Mn 2ϩ whereas the affinity for GTP is slightly greater in the presence of Mn 2ϩ . Therefore, because the rate of exchange of GDP by GTP must be essentially the same in the presence of Mn 2ϩ and Mg 2ϩ , the enhanced rate of GTP hydrolysis with Mn 2ϩ may be because of either a decreased off-rate constant for GTP binding under multiple turnover conditions or a different involvement chemically of the metal ion in the reaction. Because the kinetics of product release do not influence the single turnover results, we attribute the enhanced intrinsic GTPase activity of the Mn 2ϩenzyme to these same factors. Fig. 2 shows that the intrinsic GTPase activity of the enzyme in the absence of added divalent metal ion is unaltered from that in the presence of Mg 2ϩ . The binding of GDP in the absence of divalent metal ion is significantly weaker than in the presence of Mg 2ϩ or Mn 2ϩ , allowing even faster exchange of GDP by GTP under multiple turnover conditions. Because the intrinsic GTPase activity in the absence of added divalent metal ion is unaltered under single turnover conditions, despite a lower affinity for GTP, the hydrolytic activity in multiple turnover experiments must be considered as only apparently enhanced over that of the Mg 2ϩenzyme because of faster nucleotide exchange.

DISCUSSION
The GTPase activity of EF-Tu of T. thermophilus depleted of bound nucleotide and divalent metal ion has not been previously investigated in the absence of added divalent metal ions. The main observation, as demonstrated through Figs. 2 and 3 and Table II, is that the intrinsic GTPase activity and the kirromycin-stimulated GTPase activity of T. thermophilus 4 We have observed by Scatchard analyses that the binding of nucleotide by EF-Tu f in the presence of VO 2ϩ and 0.15 M NH 4 Cl results in values of dissociation constants equivalent to those observed in the absence of added divalent metal ion (A. Banerjee and M. W. Makinen, unpublished observations). 5 Because it was likely that VO 2ϩ would have additional binding interactions with kirromycin, different from those of the other divalent metal ions studied, and because we had limited quantities of the antibiotic, we did not evaluate kirromycin-stimulated GTPase activity in the presence of VO 2ϩ .  Table II.  (11,46), the rate-limiting step cannot be the same as for the physiological reaction of EF-Tu-catalyzed GTP hydrolysis in the presence of programmed ribosomes. If the change in rate-limiting step in the presence of programmed ribosomes is a consequence of direct participation of Mg 2ϩ in catalysis, structural relationships of the divalent metal ion and the triphosphate moiety of GTP in the active site of the EF-Tu⅐Mg 2ϩ ⅐GTP complex must differ significantly from those described heretofore (5,6). Thus, the results of our studies indicate not only that our understanding of the structural and chemical role of Mg 2ϩ in the biological function of EF-Tu is incomplete but also that the physiological relevance of the x-ray structure of the EF-Tu⅐Mg 2ϩ ⅐guanosine 5Ј-(␤,␥-imido)-triphosphate complex (5,6), prepared and crystallized in the absence of ribosomes through methods comparable with those applied here, is unclear. The differential influence of divalent metal ions on GTPase activity and the nucleotide binding affinity can be analyzed on the basis of thermodynamic relationships in Scheme 1. For the closed system of reversible reactions constructed in Scheme 1 for hydrolysis of the free Me 2ϩ ⅐GTP complex and, correspondingly, of the EF-Tu⅐Me 2ϩ ⅐GTP complex, the reactions for GTP and GDP binding and release run along the vertical direction whereas the hydrolysis of free and EF-Tu-bound Me 2ϩ ⅐GTP are represented horizontally. The protein-bound and free states of the Me 2ϩ ⅐GTP and Me 2ϩ ⅐GDP complexes are coupled through equilibria for the binding and release of each Me 2ϩ -nucleotide complex by EF-Tu. The free energy change associated with binding of each metal ion-nucleotide complex is estimated on the basis of the equilibrium (dissociation) constants in Table I. As the best approximation of the free energy change for hydrolysis of the free Me 2ϩ ⅐GTP complex, we apply the values determined through calorimetric and titration studies of George and co-workers (27, 28) for adenosine 5Ј-triphosphate under conditions of pH, ionic strength, and Mg 2ϩ concentration corresponding to those in our investigations. Because the closed series of linked equilibria in Scheme 1 requires that ⌺ ⌬G i ϭ 0, the influence of divalent metal ions on the EF-Tu-catalyzed hydrolysis of the Me 2ϩ ⅐GTP complex with respect to the corresponding hydrolytic reaction of the free Me 2ϩ ⅐GTP complex in solution can be quantified. This analysis assumes that the free energy change for hydrolysis of each Me 2ϩ ⅐GTP complex in its free form can be approximated by that for Mg 2ϩ ⅐GTP. Whereas the analysis is rigorously applied in the case of Mg 2ϩ , we anticipate that the error attributed to this assumption for the other divalent metal ions, particularly Mn 2ϩ , is not significant.
With respect to the ⌬G°Ј of Ϫ34.9 kJ/mol for hydrolysis of free Mg 2ϩ ⅐GTP in solution, the results in Table III show that the differential binding affinity of GDP and GTP by EF-Tu f in the presence of divalent metal ions results in an increasing thermodynamic driving force to stabilize the EF-Tu⅐Me 2ϩ ⅐GDP complex according to the series VO 2ϩ Ͻ Mn 2ϩ Ͻ Mg 2ϩ . Whereas Mg 2ϩ retards hydrolysis of GTP in the absence of protein, a process that underlies the stability of nucleoside 5Ј-triphosphates in the cell, Mg 2ϩ favors formation of the EF-Tu⅐Mg 2ϩ ⅐GDP complex by ϳ10 kJ/mol over the free Mg 2ϩ ⅐GDP complex under comparable conditions of ionic strength, pH, and Mg 2ϩ concentration. Furthermore, it is clear that the protein alone is not responsible for this action because EF-Tu f is not associated with differential binding affinity for GTP and GDP in the absence of divalent metal ion. Also, the thermodynamic driving force of Mg 2ϩ to stabilize the EF-Tu⅐Mg 2ϩ ⅐GDP complex is independent of the catalytic role that Mg 2ϩ may have in the physiological GTPase activity of EF-Tu in the presence of programmed ribosomes.
The analysis through Table III and Scheme 1 affirms the generally held notion that the role of Mg 2ϩ is to stabilize the

Role of Mg 2ϩ in EF-Tu Action
EF-Tu⅐Mg 2ϩ ⅐GDP complex. This protein-nucleotide complex in turn serves as a pivotal point in the regulation of the activity cycle of EF-Tu. Because the cellular content of GTP is at least 7-fold greater than that of GDP (47), EF-Tu-bound GTP upon hydrolysis in the cell would lead to saturation of the protein as the tightly bound EF-Tu⅐Mg 2ϩ ⅐GDP complex simply because of the ubiquitous presence of Mg 2ϩ . Thus, the thermodynamic role of Mg 2ϩ in EF-Tu catalyzed GTP hydrolysis requires the presence of a nucleotide exchange factor in the cell to kinetically facilitate replacement of GDP by GTP. Comparison of the x-ray structures of the GTP (5, 6) and GDP (48,49) bound forms of EF-Tu shows significant structural changes involving not only local changes in the metal ion and nucleotide binding sites but also global changes involving shifts of domains II and III. The binding surfaces on EF-Tu involved in binding to EF-Ts are located in the G domain and domain III, and their arrangement remains similar upon formation of the EF-Tu⅐EF-Ts complex to that in the EF-Tu⅐Mg 2ϩ ⅐GDP complex. It is thought that this structural relationship underlies the kinetic preference of EF-Ts to bind to the GDP complex of EF-Tu over the GTP complex (50). On this basis, it is likely that the role of Mg 2ϩ in stabilizing the GDP complex facilitates recognition of the EF-Tu⅐Mg 2ϩ ⅐GDP complex by EF-Ts. It is of interest to note that binding of T. thermophilus EF-Tu by the nucleotide exchange factor EF-Ts results in insertion of the Phe-82 residue of EF-Ts into the hydrophobic pocket of EF-Tu formed by His-85, Leu-122, and His-119, disturbing the Mg 2ϩ -phosphate portion of the active site (50), and the occupancy of Mg 2ϩ in the binding pocket, as established by x-ray studies of the Tu⅐Ts complex (50,51), is abolished. Thus, in the biological action of EF-Tu, Mg 2ϩ binding stabilizes the conformation of the EF-Tu⅐Mg 2ϩ ⅐GDP complex to facilitate recognition by EF-Ts, but destabilization of Mg 2ϩ binding in the active site of Tu is likely to be critical to initiate nucleotide exchange.