Characterization of the GTPase Reaction of Elongation Factor Tu DETERMINATION OF THE STEREOCHEMICAL COURSE IN THE PRESENCE OF ANTIBIOTIC X5108*

The stereochemical course of the GTPase of elongation factor Tu from Escherichia coli has been deter- mined by making use of the reaction dependent on antibiotic X5108 (an N-methylated derivative of kirro-mycin). Guanosine 5’-(y-thio)triphosphate stereospecif- ically labeled with I7O and “0 in the y-position was hydrolyzed in the presence of elongation factor Tu and X5108. The configuration of the product, inorganic [“jO, “0, ‘SO]thiophosphate was analyzed by 31P NMR after its stereospecific incorporation into adenosine 5’-@-thi0)triphosphate. The analysis showed that the hy- drolysis proceeds with inversion of configuration at the transferred phosphorus, implying that there is not a phosphoenzyme intermediate. Incubation of [PY-~’O, y-“03]GTP with elongation factor Tu leaves the “O-la- beling unaltered, as shown by 31P NMR. No exchange of oxygens with water nor py-p positional isotope ex- change occurs, implying that not even transient cleavage can occur with the elongation factor alone. Only on interaction with X5108, kirromycin, or ribosomes does the cleavage occur, most likely by a single step, in-line transfer of the terminal phosphorus from GPP to a water oxygen. These properties ofthe GTP hydrolysis mechanism of elongation factor Tu are similar to those of elongation factor G. In the elongation


S (retention) R (inversion)
Whether this reaction occurs with retention or inversion of configuration can be determined (Webb and Trentham, 1980a;Webb, 1982). There is strong evidence that the stereochemical courses of thiophospho and phospho transfer are identical (Eckstein et at., 1982), so that the result with GTPyS should give information as to whether or not GTP hydrolysis occurs via a phosphorylated intermediate, such as a phosphoenzyme. Each phosphotransfer step probably occurs with inversion, so that overall inversion implies a single step via a direct, in-line mechanism (Knowles, 1980). Overall retention is explained most simply by two steps: the formation and breakdown of a phosphorylated intermediate. This stereochemical technique has been applied previously to three ATPases, myosin, mitochondrial and sarcoplasmic reticulum, and the results support this explanation for triphosphatases (Webb andTrentham, 1980b, 1981;Webb et al., 1980). Thompson and Karim (1982) have shown that a ternary complex of Phe-tRNA. EF-Tu. GTP@ binds to poly(U)-programmed ribosomes. The subsequent hydrolysis of GTPyS is analogous to that of GTP, but it occurs at a rate at least 2500 times slower. This physiological reaction is therefore not suitable in practice for the stereochemical determination, since the amounts of reactants required would be much too great. However, the antibiotic kirromycin interacts with the protein EF-Tu to produce several effects on the elongation system. One effect is that it can induce GTPase activity in EF-Tu, uncoupled from interaction with ribosomes. Chinali et al. (1977) have shown that this GTP hydrolysis shares common features with the physiological reaction, so that it is probable that the two reactions share a common active site and chemical mechanism. In this paper, we make use of this uncoupled hydrolysis to determine the stereochemical course of EF-Tu-catalyzed GTPase activity, in the presence of antibiotic X5108, an N-methylated derivative of kirromycin. X5108 binds tightly to EF-Tu (Wilson and Cohn, 1977;Eccleston, 1981a) and has similar effects as kirromycin itself on the partial reactions of EF-Tu .
Using oxygen exchange methods, we have investigated the possibility that the cleavage step of the EF-Tu GTPase mechanism is reversible. The results are compared with those obtained with EF-G, the elongation factor involved in the translocation process of the peptide elongation cycle.

Stereochemistry of Elongation Factor
EF-TU. GTP from Escherichia coli was prepared and characterized as described by Eccleston (1981b). EF-G and ribosomes from E . coli were prepared as described by Webb and Eccleston (1981). Antibiotic X5108 was from Hoffmann-LaRoche and used as a 25 mM solution in ethanol. ($3) r~y -'~O ; y"0, "OIGTPyS was synthesized and characterized by the methods described by Webb and Eccleston (1981). Guanylate kinase from bovine brain and nucleosidediphosphate kinase from bakers' yeast were from Sigma.
The conditions used to hydrolyze GTPyS in the presence of EF-Tu and X5108 were based on those used by Sander et al. (1979). A solution containing 20 mM Tris/HCl, pH 7.5, 10 mM MgC12, 2 mM dithiothreitol, 100 mM NH4C1, 100 p~ EF-Tu-GDP, 500 PM GTPyS, 250 p~ X5108, 0.2 units.&" of guanylate kinase, 5 units.ml-' of nucleosidediphosphate kinase, and 200 p~ ADP was incubated at 37 "C. Samples were taken at intervals and the extent of hydrolysis was determined by high performance liquid chromatography as described by Webb and Eccleston (1981). In addition, formation of inorganic thiophosphate was determined by high performance liquid chromatography with postcolumn derivitization. The chromatography used a Waters system and a Whatman Partisil 10/25 SAX ion exchange column (25 cm x 4.6 mm diameter). The material was eluted with 0.25 M ammonium phosphate, pH 4.0, at 0.8 m1.min-I. The column eluate was mixed, using a Valco zero volume-T connection, with 1 M Tris/HCl, pH 8.0, containing 1 mM 5,5'-dithiobis-(2-nitrobenzoic acid), also at a flow rate of 0.8 ml.min", maintained by a second Waters 6000A pump. The mixed solution was passed through a 1-ml delay tube before detection at 412 nm.
ATPPS as described by Webb (1982), except that all reactions and a l l Inorganic thiophosphate was incorporated stereospecifically into elution profiles of ion exchange columns were followed by high performance liquid chromatography. The inorganic thiophosphate was purified by DEAE-cellulose chromatography and the product was detected as described above. Each of the thiophosphoryl nucleotides formed during the incorporation were detected and quantitated by high performance liquid chromatography using conditions described for ATPyS (Webb, 1982). This enabled the reactions to be quenched when the product formation was maximal so reducing possible isotope losses. The final product, ATPPS, was analyzed by 31P NMR (Webb and Trentham, 1980a;Webb, 1982).
The ability of EF-Tu to catalyze oxygen exchange in the absence of antibiotic or ribosomes was tested using [Py-"O; y-"00olGTP, enriched to an extent of 978 in each labeled position. This GTP was synthesized and the I8O distribution about its , I 3 and y phosphorus atoms was determined by "P NMR as previously described (Webb, 1980). Nucleotide-free EF-Tu was prepared as described by Eccleston et al. (1981), except that EDTA was not removed. It contained 20% residual GDP. This EF-Tu (at 50 p~) was incubated with 0.5 mM labeled GTP for 4 h at 0 "C, in 50 mM Tris/HCl, pH 7.5, 12 mm MgCL, 2 mM EDTA. High performance liquid chromatography showed no hydrolysis occurred. The GTP was purified by ion exchange chromatography and analyzed by "P NMR as previously described (Webb, 1980). A similar experiment was done with EF-G. The following solution was incubated for 2 h at 30 "C: 0.12 mg.ml" of EF-G, 1 mM [Py-"O; y-'*Os]GTP, 20 mM Tris/HCl, pH 7.8, 15 mM MgC12, 1 mM dithiothreitol, and 80 mM NH4C1. Essentially no hydrolysis occurred during this time. The GTP was analyzed as described above and there was no change in the pattern or extent of labeling.
The extent of intermediate oxygen exchange with water during GTP hydrolysis in the presence of EF-G and ribosomes was determined using [y-''O:j]GTP. Three different ribosome concentrations were used, for which the rates of hydrolysis differ widely. [y-180:l]GTP (Webb and Trentham, 1981b), 95% enriched at each labeled position, was incubated at 2 mM concentration with 0.1 mg-ml" of EF-G at 30 "C in a solution containing 20 mM Tris/HCl, pH 7.8, 15 mM MgCla, 1 mM dithiothreitol, and 80 mM NH4CI. The extent of hydrolysis in the presence of ribosomes (at 0.06, 0.23, and 0.92 p~) was determined by high performance liquid chromatography. The hydrolysis rates were 0.15, 0.50, and 1.33 s-I, respectively, compared with 0.005 s-I in the absence of ribosomes. Measurements at even higher ribosome concentrations showed that maximal activation had not been achieved at 0.92 PM ribosomes. These rates are comparable with those observed by Rohrbach et aE. (1974). After 90% hydrolysis, the P, was purified by ion exchange chromatography and analyzed for by "'P NMR. All three samples showed the same enrichment as the starting GTP. steady state GTPyS hydrolysis rates Solutions at 37 "C contained 20 mM Tris/HCl, pH 7.5, 10 mM MgCla, 2 mM dithiothreitol, 100 mM NH4CI, and 0.5 mM GTPyS. The coupling system was 200 p~ ADP, 0.2 units. ml" of guanylate kinase, and 5 units.&" of nucleoside diphosphate kinase. The concentrations of itemized additions were 100 PM EF-Tu and 250 PM X5108. Samples (10 pl) were withdrawn at intervals and the disappearance of GTP+ was analyzed by high performance liquid chromatography as described in the text. The rate of hydrolysis was calculated from the linear portion of the reaction time course.

RESULTS AND DISCUSSION
In order to determine that the only GTPase activity in our EF-Tu preparation was the reaction dependent on X5108, the rates of G T P hydrolysis were measured in the absence and presence of the antibiotic. In the absence of X5108, there was no observable hydrolysis (t33, ~L M h"). In its presence, the rate was approximately 1000 p -h -l . This is equivalent to a rate constant of 0.03 s-I, so that hydrolysis is limited probably by the rate of GDP release from EF-Tu, consistent with the results of Fasano et al. (1978).
GTPyS hydrolyzes slowly in the presence of EF-Tu and antibiotic X5108, as shown in Table I. There is also a significant background breakdown in the absence of the protein, consistent with the instability of GTPyS solutions." The rate of breakdown in the presence of EF-Tu but without X5108 was similar to the background rate. The products of these background reactions were not determined, although it is probable that, at least in part, the GTPyS is destroyed by loss of sulfur. With X5108, the rate of disappearance of GTPyS is 4-fold larger in the presence of EF-Tu than in its absence. The products in this case were shown by high performance liquid chromatography to be GDP and inorganic thiophosphate. However in the absence of the coupling system consisting of ADP, guanylate kinase, and nucleoside diphosphate kinase, which converts GDP to GMP, the rate of hydrolysis was consistently slower than in its presence and the hydrolysis further slowed down after partial reaction. This is probably due to inhibition by GDP (Miller and Weissbach, 1970). Hence the coupling system was added in all reactions in Table I, except where indicated, so that there was no significant GDP accumulation. GMP forms and binds at least lo4 times more weakly to EF-Tu than does GDP (Miller and Weissbach, 1977).
These test reactions show that at least 60% of the GMP + inorganic thiophosphate was due to the X5108-dependent GTPase of EF-Tu. For the stereochemical determination, the reaction was done on a scale of 20 ml, using GTPyS stereospecifically labeled with ' .' O and '"0. After a reaction time of whether the hydrolysis proceeded with retention or inversion of configuration. Its configuration was determined after incorporation into ATPpS by 31P NMR of the P-phosphorus (Webb and Trentham, 1980a). The formation of ATPpS is highly stereoselective, but one oxygen is lost from the thiophosphate during the incorporation. One third of the molecules lose '"0, one third lose 1 7 0 , and one third lose "0. Molecules that retain I7O are not visible in the NMR spectrum due to quadrupole broadening of the /I-phosphorus signal by the nuclear spin of 170. Molecules that lose 1 7 0 are visible and give rise to ATPpS containing an "0 which is nonbridging (species 1) or bridging (2) depending on whether it is derived from S or R inorganic thiophosphate:

R 2
These two ATP@ species are distinguished by "'P NMR, due to slightly different chemical shifts (Webb and Trentham, 1980a). Isotopic enrichments are less than 100% in the GTPyS so that there are other inorganic thiophosphate species, partially labeled and unlabeled, apart from the l 6 0 , "0, "0-labeled, chiral species. These give rise to other peaks in the NMR spectrum of ATPPS apart from those due to species 1 or 2. This spectrum is shown in Fig. 1, in which the peaks are assigned to the various ATP@ species as described by Webb and Trentham (1980a). Qualitatively the excess ATPDS with bridging "0 over that with nonbridging "0 shows that the major pathway for GTPyS hydrolysis is inversion. Table I1 shows the relative peak intensities from this spectrum, compared to those calculated for inversion or retention. Once the lack of isotopic purity of ATPpS is taken into account, the spectrum is close to that calculated for all GTPyS hydrolysis occurring with inversion.
It is probable that the hydrolysis of GTP, catalyzed by EF-Tu, has the same chemical mechanism as GTPyS and so occurs with inversion of configuration at the transferred phosphorus atom. This is strong evidence against a phosphorylated intermediate and is consistent with the reaction occurring in a single step, with direct, in-line displacement of GDP by a water oxygen.
This mechanism of the GTPase was investigated further by oxygen isotope experiments to detect transient cleavage of GTP on EF-Tu. Although EF-Tu does not catalyze significant hydrolysis of GTP in the absence of ribosomes or kirromycin, the possibility remains that reversible hydrolysis occurs as in Equation 3 but that no product is released.
EF-Tu .GTP e EF-Tu . GDP. P, If this were the case, the EF-Tu system would parallel actomyosin, since myosin catalyzes the rapid, reversible cleavage of ATP even in the absence of actin (Taylor, 1979). To detect such transient, reversible cleavage we used G T P labeled with I8O as in species 3: When this GTP was incubated with EF-Tu for 4 h at 0 "C, no change occurred to the pattern or extent of labeling, as determined by 31P NMR. This implies that no GTP is cleaved reversibly on the protein: that is, it is not cleaved transiently to EF-Tu. GDP.Pi then re-formed to EF-Tu .GTP. If this transient reaction with water occurred, the labeling would have changed probably in two ways, shown in species 4 of Equation 4. In one process, oxygen exchange with the water would have occurred. This exchange would occur on transient cleavage, since the first-formed Pi is P. 30. On re-forming GTP, there is a chance that "0 will be lost from the molecule resulting in only partial "0 enrichment (a) remaining in the y-phosphate. Also, the transient cleavage would probably give rise to positional isotope exchange (Midelfort and Rose, 1976). The GTP would be cleaved between the /3y-'*O and the y-P. The I8O remaining on GDP would become equivalent with the other P-oxygens ("0). On re-forming GTP, one of these p-  " These spectra are calculated from the known isotopic enrichments of the GTPyS: 85% for I8O, 45% for "0. The I7O positions also contain 30% "0. The calculation assumes a 9% loss of isotope during the hydrolysis of GTPyS or during incorporation of inorganic thiophosphate into ATPPS, as described by Webb et al. (1980), to obtain the observed '*O-enrichment in ATPPS. No attempt was made to determine the extent of exchange during hydrolysis of GTPyS since all such determinations in other systems (Webb, 1982) have found negligible extents of exchange. The noise in the spectrum causes an uncertainty of approximately &I% in the observed peak intensities.
oxygens ( 1 6 0 2 1801) of GDP becomes the ,@-bridging oxygen of GTP. This bridging oxygen, therefore, would be l60 in some molecules. Both oxygen exchange with water and positional isotope exchange can be detected by the change in '*O labeling measured by 31P NMR. Since no changes from 3 to 4 occurred, it is probable that no reversible, transient cleavage o c c~r r e d .~ The cleavage is only facilitated on interaction with ribosomes or kirromycin, when the direct displacement of GDP from G T P by water occurs. This behavior is comparable to that of EF-G, which also catalyzes G T P hydrolysis (in the presence of ribosomes) with inversion of configuration . It also does not seem to catalyze even transient cleavage of G T P in the absence of ribosomes since a similar experiment with 180-labeled GTP and EF-G, as described under "Experimental Procedures," showed no change in the pattern or extent of labeling. It has been shown also that there is no intermediate oxygen exchange with water during G T P hydrolysis in the presence of EF-Tu (from Bacillus stearothermophilus) and XE1108.~ This is most likely due to the cleavage being irreversible although product release can occur. A similar lack of exchange with water has been observed for GTP hydrolysis in the presence of EF-G and a saturating concentration of ribosomes (Rohrbach et al., 1974). In this case, the rate of hydrolysis increases with increasing ribosome concentrations. However, no exchange occurs over a wide range of activation as described under "Experimental Procedures." Thus, the cleavage is irreversible even at low activation, when the velocity of product release is low.
There is therefore, a close similarity in the mechanism of the uncoupled GTPase of the two elongation factors. Both catalyze the hydrolysis with inversion of configuration and the cleavage appears to be irreversible. It is probable that the cleavage cannot occur in the absence of activating kirromycin or ribosomes. This description contrasts with that of myosin, which catalyzes the rapid, reversible cleavage of ATP, with inversion of configuration, in the absence of actin. The overall hydrolysis rate is increased by the presence of actin, which increases the rate of product release, but has only a small effect on the rate of cleavage (Stein et al., 1981). Cleavage becomes effectively irreversible only a t high actin concentra-This is the most likely explanation of the result. It is possible that cleavage occurs but that all / 3 and y oxygens are held rigid (for example, by tight metal coordination) so that the phosphates cannot rotate. If such rotation cannot occur, the oxygens cannot exchange positions.
' J. S. Taylor and M. Cohn, personal communication. tions. It is not yet understood how these differences in reversibility of the cleavage step of the two energy-transducing systems (the cross-bridge of muscle and the elongation cycle of protein biosynthesis) are related to the biological function. However, nucleotide binding and cleavage clearly have large effects on the protein-protein interactions in both systems.